Slashdot Mirror


California Legalizes Self Driving Cars

Hugh Pickens writes writes "The Seattle PI reports that California has become the third state to explicitly legalize driverless vehicles, setting the stage for computers to take the wheel along the state's highways and roads ... 'Today we're looking at science fiction becoming tomorrow's reality,' said Gov. Brown. 'This self-driving car is another step forward in this long, march of California pioneering the future and leading not just the country, but the whole world.' The law immediately allows for testing of the vehicles on public roadways, so long as properly licensed drivers are seated at the wheel and able to take over. It also lays out a roadmap for manufacturers to seek permits from the DMV to build and sell driverless cars to consumers. Bryant Walker Smith, a fellow at Stanford's Center for Automotive Research points to a statistical basis for safety that the DMV might consider as it begins to develop standards: 'Google's cars would need to drive themselves (by themselves) more than 725,000 representative miles without incident for us to say with 99 percent confidence that they crash less frequently than conventional cars. If we look only at fatal crashes, this minimum skyrockets to 300 million miles. To my knowledge, Google has yet to reach these milestones.'"

508 comments

  1. Must past this test by o5770 · · Score: 3, Interesting

    Here is a scenario where if a self-driving car can pass 100% of the time, then I would deem it safe to get into.

    Driving on a mountain road around a sharp corner where there is a steep cliff on the right side. Auto-car is passed on the left by some *sshole "manual" driver, but then the *sshat driver cuts in short because of oncoming traffic at the last second. Robo-driver identifies there is suddenly a car intruding into its safe-T-zone (TM) and does what its programming tells it to do, avoid hitting other vehicles. So the self-driving wonder swerves right to avoid the other car and zooms off the cliff.

    A human driver would recognize that hitting the other car in this instance is the safer solution then to go careening off the steep cliff.

    I agree that a self-driving car can work, and 99% of the time will perform adequately to protect its occupants from disaster. But since we have not mastered true AI yet, all self-driven cars will be built with flaws in their logic that will fail catastrophically. "Avoid hitting all cars", for instance, is not a good enough directive to ensure the safety of the occupants in 100% of all situations.

    Someone mentioned that the deaths caused by self-driven cars would be far less then manual drivers, but then I would disagree that any technology introduced on the highways would be adequate to allow any fatality, especially in scenarios where a human driver may have been able to avoid death.

    Basically what I am waiting for is the inevitable 100 car pile up with massive fatalities that WILL occur at some point in time where investigation will identify that a self-driven car, or cars, was the cause of it. Any company involved in programming or manufacturing that self-driven car will be sued out of existence and the "love affair" everyone seems to have about auto-driving cars will end quickly.

    I am amazed at how delusional governments are into so quickly allowing this technology on the roads, sounds to me like there is some massive lobbying going on to short-cut the necessary amount of time to test auto-driven cars under all senarios, not just ones in controlled and predictable setups like we have seen. 5 years ago robo-cars could not drive around a dirt track, now they are quickly being allowed on our highways. That just is irresponsible.

    1. Re:Must past this test by Anonymous Coward · · Score: 5, Insightful

      A human won't pass that test 100% of the time either, so I'm not sure what your point is about 100%. It's all statistics.

    2. Re:Must past this test by h4rr4r · · Score: 4, Insightful

      Why would a self driving car ever drive off a cliff?
      Clearly it would rank available options and pick the lowest cost one. The cheapest collision in that case.

      Human drivers allow fatalities everyday. The question is not is it better than some hypothetical human driver, but is it better than the drivers we have right now.

      5 years ago the tech to do this was not cheap enough, now it is. This is called progress not being irresponsible. What is irresponsible is suggesting that the average person continue to drive automobiles when we have a better solution at hand.

    3. Re:Must past this test by queazocotal · · Score: 3, Interesting

      I note that in the USA, the pass rate of the driving test in general exceeds 50% by a considerable margin.
      This is not due to great tuition and driver skill and knowledge.
      Also, a number of other safety features that would considerably reduce deaths are not implemented.

      If the autodriver is safer than the average auto driver, ...

    4. Re:Must past this test by flyingfsck · · Score: 1

      That may be a problem in the lawsuit happy USA, but in the rest of the world, self driving cars will improve by leaps and bounds. Anyhoo, a self driving car crashing, is an industrial accident, there are already laws for that.

      --
      Excuse me, but please get off my Pennisetum Clandestinum, eh!
    5. Re:Must past this test by cyberchondriac · · Score: 1, Insightful

      My first gut instinct is, this is bad, bad, bad.. but then I think of the stupid beatch in the Hyundai that blew by me at 85mph, then cut into my lane, making me slam my brakes on while driving to work this morning.. so maybe it's not so bad.

      --

      Look back up at my post, now look back down, you're on the Internet. Now look back up. I'm a signature.
    6. Re:Must past this test by rich_hudds · · Score: 4, Insightful

      I think you're entirely wrong.

      A much more likely scenario is that the self driving cars prove statistically to be safer than human driven cars.

      At that point expect legislation to ban humans from driving.

      Imagine trying to defend yourself in court if you've caused a fatal accident.

      'Why did you turn off the computer when you know it is proven to be safer?'

    7. Re:Must past this test by Anonymous Coward · · Score: 1

      First, you seem to be assuming the current cars would not pass this test.

      Second, you're assuming all driving situations are appropriate for these vehicles, e.g. winding mountain roads as opposed to straight, flat highways.

      Third, as has been pointed out, what is the current pass rate on this test for human drivers?

    8. Re:Must past this test by trancemission · · Score: 1

      As you seem to like statistics, I don't think the cars will be 'self driving' 100% of the time. If I was 'driving' the car and came accross a steep drop - I would take control.

      Are you expecting to just hop in your car and fall asleep and let your car take you to your destination?

      You might aswell keep dreaming for flying cars all over the place...

    9. Re:Must past this test by BasilBrush · · Score: 2

      The plan to allow test vehicles to cover a large number of miles and then compare collision/fatality stats with human drivers is the correct one. It's quite likely that the auto-driver will make different mistakes than the typical human driver. For the sake of argument, suppose it has a greater tendancy to make the mistake you outline than a human driver does. That doesn't matter if it also avoids more collisions and fatalities in other scenarios. If the stats say you get fewer collisions and fatalities with auto-drivers, then it would be madness not to allow them.

    10. Re:Must past this test by NeutronCowboy · · Score: 5, Interesting

      So the self-driving wonder swerves right to avoid the other car and zooms off the cliff. A human driver would recognize that hitting the other car in this instance is the safer solution then to go careening off the steep cliff.

      Someone has never, ever taken an AI class. Or even an algorithm class dealing with risk. Here's how the calculation actually works (and by the way, that approach is about 20-30 years old).
      Every situation is assessed an impact value: driving into oncoming traffic, 0 (very bad); driving into the right ditch, 10; swerving into a legal lane, 50; etc. Every situation is given a set of possible actions, with each action having a probability of being completed successfully. The algorithm multiplies the outcome with the odds of achieving that outcome, and picks the highest value. You can set it up in different ways, but the idea is the same: multiply outcome severity with odds of achieving outcome, pick lowest combined risk/outcome. In your situation, driving off the cliff (which is assumed to be very bad, since the car can see a very steep drop-off with no bottom) is going to have a much worse outcome than hitting the car in front of it. Hitting the car in front of it is guaranteed, but so is driving off the cliff. As a result, the algorithm will make the automated car hit the car in front of it, rather than drive off the cliff.

      Not to mention that cars don't sleep, always behave optimally (according to the algorithms in place), and have no blind spots.

      Basically what I am waiting for is the inevitable 100 car pile up with massive fatalities that WILL occur at some point in time where investigation will identify that a self-driven car, or cars, was the cause of it.

      You mean like the ones that regularly happen in fog and icy/rainy conditions?

      Any company involved in programming or manufacturing that self-driven car will be sued out of existence and the "love affair" everyone seems to have about auto-driving cars will end quickly.

      That is a very real risk. Not sure how the laws will deal with it. But until that question is addressed, we won't see large-scale sales of automated cars. I suspect that we'll see the equivalent of ToS: by using this car, you agree to be fully responsible for all its actions and accidents.

      --
      Those who can, do. Those who can't, sue.
    11. Re:Must past this test by h4rr4r · · Score: 3, Interesting

      We can only hope that driving tests become harder and harder and only those who pass them will be allowed to drive themselves.

      Why would you ever want to turn off the automated driver? Do you think rich folks are constantly putting their limo driver in the back and taking the wheel themselves?

    12. Re:Must past this test by h4rr4r · · Score: 1

      Why would you take over?
      This is a task a machine will always beat you at, it depends mostly on reaction time.

      Yes, I expect to have my car come pick me up from the bar and drive my drunk self home. Otherwise this is pointless.

    13. Re:Must past this test by CanHasDIY · · Score: 2, Insightful

      'Why did you turn off the computer when you know it is proven to be safer?'

      "Because my brain operates at a frequency modern computers cannot even begin to match, and it cannot be hacked."

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    14. Re:Must past this test by queazocotal · · Score: 1

      You missed a cost.
      thisAlgorithmBecomingSkynetCost=-999999999

    15. Re:Must past this test by BasilBrush · · Score: 2

      If I was 'driving' the car and came accross a steep drop - I would take control.

      Why? It's not an unexpected obstacle. By the time autodrivers are allowed in consumer cars you can expect the auto-driver to be able to deal with it better than you can.

      Are you expecting to just hop in your car and fall asleep and let your car take you to your destination?

      Yes. I see no reason why at some point in the future an auto-driver will be a statistically safer driver than I am. So why not?

    16. Re:Must past this test by h4rr4r · · Score: 5, Insightful

      1. your reaction time is absolute crap.
      2. advertisers disagree with your notion that human brains cannot be hacked.

    17. Re:Must past this test by BaronAaron · · Score: 3

      The system just needs a rapid manual override and a little common sense from the driver.

      I see self driving cars as an evolution of cruise control. Just as cruise control gets out of your way as soon as you manually press the accelerator or brake the auto drive system should get out of your way as soon as you move the steering wheel.

      Also, drivers should take responsibility when they feel it's safe to engage the auto drive. I wouldn't use cruise control on a narrow mountain road, neither would I use auto drive. I would love to be able kick on auto drive on a long boring highway though and focus on a phone call or whatever.

    18. Re:Must past this test by cyberchondriac · · Score: 1

      No, you read it a little too literally, what I gave was the gist of the situation: first, she blew by me, then couldn't proceed any further because she quickly came up to the back bumper of a car in her lane that was going slower than everyone else, so she had to slow down; then she decided my lane was better..so she cut me off as she squeezed over between me and the car in front of me, when there definitely wasn't a safe amount of space to do that. Jesus, did you want a second by second account with video too?

      --

      Look back up at my post, now look back down, you're on the Internet. Now look back up. I'm a signature.
    19. Re:Must past this test by OzPeter · · Score: 2

      If I was 'driving' the car and came accross a steep drop - I would take control.

      As someone mentioned below .. if your "driverless car" experience is to sit there waiting to take control of the device when you sense that it is about to get into trouble, then that is going to be a stressful and shitty experience. You might as well have been driving yourself all of the time.

      --
      I am Slashdot. Are you Slashdot as well?
    20. Re:Must past this test by Altanar · · Score: 4, Interesting

      Self driving cars *never* swerve. They brake. Statistically they know that swerving almost always is worse than the incoming accident. Humans on the other hand will swerve. See all the accidents that occur when attempting to miss an animal crossing the road.

    21. Re:Must past this test by Anonymous Coward · · Score: 1

      Incorrect, your brain operates at somewhere around 100hz, it's just massively parallel in a scale we cannot even begin to match

      *yet^

      ^Providing you don't count large supercomputing projects that can simulate smallish fractions of the human brain.

    22. Re:Must past this test by Anonymous Coward · · Score: 0

      he's one of those assholes that goes 45 on the highway, but speeds up the moment anyone tries to pass him. So the other car ends up getting to 85 just to get around him, and then he has to slam on his brakes to avoid hitting her.

    23. Re:Must past this test by MyLongNickName · · Score: 1

      I don't think it is a question of the algorithm, but rather a question on the computer's ability to recognize the situation accurately. Machine vision has improved a lot, but is it to the point that it can recognize all the situations brought up in the OP. Maybe it can, but I think the only real test is extended real world testing.

      --
      See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
    24. Re:Must past this test by Anonymous Coward · · Score: 0

      Well there might be situations like a cliff that aren't handled better than a human by the program. In those cases you could mark the road on gps and force the human to take control if they want to take that route.

    25. Re:Must past this test by Anonymous Coward · · Score: 0

      That is a very real risk. Not sure how the laws will deal with it. But until that question is addressed, we won't see large-scale sales of automated cars. I suspect that we'll see the equivalent of ToS: by using this car, you agree to be fully responsible for all its actions and accidents.

      No-one - with the possible exception of a few of the top engineers working on the thing, or maybe an insurance actuary who studies self-driven cars - can actually agree to that in a fully informed manner. Far better would be some kind of insurance agreement between the car manufacturer, the owner, and possibly/probably a third party insurer.

    26. Re:Must past this test by slew · · Score: 1

      ... but in the rest of the world, self driving cars will improve by leaps and bounds...

      Depending on where in the world you are, this might be a necessity. Observing the driving habits I've seen in many countries of the far east and parts of southern europe, self driving cars *better* improve by leaps and bounds just to survive!

    27. Re:Must past this test by BasilBrush · · Score: 5, Interesting

      1) You reaction time is far worse than a computer.
      2) Your estimation of distances is far worse than machines absolute measurements.
      3) You are limited to two forward facing eyes, augmented by 3 small mirrors. And you share some of the vision time with looking at the dash. An auto-car can look in all directions at once, and monitor all dashboard information and more at the same time.
      4) An auto-driver will be better at maintaining a safe speed. Able to stop in the distance it knows to be clear far more often than a human driver.
      5) I'd expect an auto-driver system to be seperate from any other computing devices in the car, and connected to the internet or any other vector for hacking. I'd expect them to be as immune to hacking as an auto-pilot system in a plane.

    28. Re:Must past this test by Mike+Buddha · · Score: 2

      Yeah, and cruise control is stupid because you can't use it all the time, too.

      If I have a hammer I have to use it for every single task I have, otherwise what's the point of owning a hammer?

      --
      by Mike Buddha -- Someday the mountain might get him, but the law never will.
    29. Re:Must past this test by 0123456 · · Score: 2

      The question is not is it better than some hypothetical human driver, but is it better than the drivers we have right now.

      No, the question is: is it better than me?

      If not, I don't want it driving my car.

    30. Re:Must past this test by MobileTatsu-NJG · · Score: 1

      Driving on a mountain road around a sharp corner where there is a steep cliff on the right side. Auto-car is passed on the left by some *sshole "manual" driver, but then the *sshat driver cuts in short because of oncoming traffic at the last second. Robo-driver identifies there is suddenly a car intruding into its safe-T-zone (TM) and does what its programming tells it to do, avoid hitting other vehicles. So the self-driving wonder swerves right to avoid the other car and zooms off the cliff.

      Wasn't this an episode of Knight Rider?

      --

      "I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)

    31. Re:Must past this test by Anonymous Coward · · Score: 0

      you ever want to turn off the automated driver?

      How about driving is fun and getting carried is not.

    32. Re:Must past this test by h4rr4r · · Score: 5, Insightful

      No that question is; Is the car a better driver than me when I am sleep deprived, upset at my wife and in a hurry to get home?

      The computer will always drive the same, humans are not the reliable.

    33. Re:Must past this test by Fauxbo · · Score: 2

      For 99% of people the subjective answer will be 'no', but the objective answer will be 'yes'.

      Guess which one will win?

    34. Re:Must past this test by Martin+Blank · · Score: 2

      For the foreseeable future, there will be times when it's necessary to disable the autodriver. New roads that aren't in the GPS system, for example, or private driving areas (e.g., parking lots) that aren't well-mapped.

      And sometimes it's just more fun to drive the car yourself.

      --
      You can never go home again... but I guess you can shop there.
    35. Re:Must past this test by Anonymous Coward · · Score: 0

      I don't think something equiped with lasers, cameras, etc, would opt to drive off a cliff... these things are brighter than most of the drivers I see on the road. How many fender benders aren't reported? how many drive-away accidents? how many "oops hit the curb/sign/garage door"? This standard is much lower than the article mentions.

      Picture the same scenario you layed out, add in the sun setting and being right in a human driver's eyes... they press the brakes in those instances, they don't rear end or drive off the road. If Car 1 stops and Car 2 stops, a computer can damn well know how much pressure needs to be applied to stop as fast and avoid other cars much better than you or I can. Best example... when a light turns green every car should theoretically be able to start moving at the same time... we know this is not true... computers are better drivers, soon you won't need to buy a car, just lease it for your errand... it will come pick you up, drop you off, do errands in between, and then drive you home. No need for driveways.

    36. Re:Must past this test by h4rr4r · · Score: 1

      Driving on a track is lots of fun, on the public roads it is very boring if you are driving safely.

    37. Re:Must past this test by h4rr4r · · Score: 2

      Way to find analogies that totally suck.

      If you have to watch the computer driving at all time and be ready to take over instantly that is torture not a method of transportation.

    38. Re:Must past this test by Anonymous Coward · · Score: 0

      Solution: outlaw human driving =)

    39. Re:Must past this test by Cro+Magnon · · Score: 1

      For the foreseeable future, there will be times when it's necessary to disable the autodriver. New roads that aren't in the GPS system, for example, or private driving areas (e.g., parking lots) that aren't well-mapped.

      And sometimes it's just more fun to drive the car yourself.

      I just replaced my Android-powered car with an iOS 6, and the maps aren't up to par yet.

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    40. Re:Must past this test by Haxagon · · Score: 1

      I accidentally modded you down, so I'm using this comment to negate it.

    41. Re:Must past this test by freeze128 · · Score: 1

      There is another solution to your scenario: The auto-car could apply the brakes. EVERYONE LIVES!

    42. Re:Must past this test by lister+king+of+smeg · · Score: 1

      no but rewatching firefly, sleeping, palying games on my tablet is fun whice unless you are an horrible driver are not thing you should be doing while driving

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    43. Re:Must past this test by Anonymous Coward · · Score: 0

      Obvious troll is obvious. High UID poster with no comment history. How did this get modded up?

    44. Re:Must past this test by Matimus · · Score: 4, Insightful
      I have known a few terrible drivers in my life. Despite their friends, and occasionally strangers, telling them that they were terrible drivers, multiple collisions in which vehicles have been totaled, and even collisions with pedestrians, they still believed that they were good drivers. Individuals may not be the best judges of whether or not they can drive better than a machine.

      It will be interesting to see how this plays out. How the public perceives it. How it is marketed. How it is handled by insurance companies.

      --
      GENERATION 25: The first time you see this, copy it into your sig on any forum and add 1 to the generation. Social exper
    45. Re:Must past this test by slew · · Score: 1

      Any company involved in programming or manufacturing that self-driven car will be sued out of existence and the "love affair" everyone seems to have about auto-driving cars will end quickly.

      That is a very real risk. Not sure how the laws will deal with it. But until that question is addressed, we won't see large-scale sales of automated cars. I suspect that we'll see the equivalent of ToS: by using this car, you agree to be fully responsible for all its actions and accidents.

      That risk is already with us. Just witness the whole debacle concerning Toyota and the drive-by-wire sticky accelerator issue. After a lengthy investigation (where they dragged in NASA and NHTSA engineers to examine thousands of lines of embeded software code), it was found that although a few of them were the result of the accelerator pedal getting wedged under a floor mat, most of the reported accidents were likely driver error (panicking and hitting the accelerator pedal instead of the brake pedal), but that didn't stop the lawsuits. Lawyers will laugh at your ToS all the way to court.

      I suspect the end game to this is that the software will be purchased from a different company that sells you the hardware. For most folks that will mean that we will not be able to purchase self-driving cars, but must rent them from integrator companies. Fortunatly, you'll be able to sleep well at the driver seat knowing that they bought the car and the software from the lowest bidder. Think a company like HP, buying from Intel and Microsoft and renting you the PC (so you can't tamper with it to install your own mods and they get to force update to fix bugs rather then let you continue to use the old buggy version).

      I also suspect that google will probably not want to sell the software to any end-users (to avoid liability), and we will get a bunch a fly-by-night self-driving software competitors vying for business from these integrator companies. Welcome to the self-driving future ;^)

    46. Re:Must past this test by Martin+Blank · · Score: 1

      And you're a self-driving car engineer? There are plenty of cases where swerving is the better choice. I've experienced it myself where while braking and swerving, I missed by maybe a foot or two another car that jumped into my lane and slammed on the brakes even though he had plenty of space in front of him. The carpool lane was fortunately empty and all that happened to me was a jolt of adrenaline, but it's one example where swerving is necessary. There are others that involve objects falling off of other vehicles and debris in the roadway that wasn't visible because of other vehicles. They're fairly rare per mile driven, but still happen often enough that swerving is a viable and useful reaction. Autodrivers will presumably be better at it because they can take into account which lanes are actually open and react like a prepared professional--or better.

      --
      You can never go home again... but I guess you can shop there.
    47. Re:Must past this test by gorzek · · Score: 2

      If the car knows there is a cliff on the right (which it should, otherwise it shouldn't be driving at all) then it will have to quickly brake and possibly hit the car in front of it. It can handle this better than a human driver in a few ways:

      1. It can gauge the right balance of braking force to minimize impact and inertia transferred to the passengers.
      2. It can pre-emptively deploy safety measures a fraction of a second sooner to protect the passengers.

      There are going to be situations where an accident is simply unavoidable. A self-driving car can make the determination on how best to minimize trauma to the passengers a lot faster than a human driver could. Cars aren't magic--they're physics, and rather simple physics, as far as that goes.

      Besides that, I find it hard to believe anyone developing these systems wouldn't try everything they know to make it screw up--and then fix its responses to any situation it didn't handle properly.

    48. Re:Must past this test by Anonymous Coward · · Score: 0

      There is another solution to your scenario: The auto-car could apply the brakes. EVERYONE LIVES!

      Except the idiot human driver who's tailgating goes into the back of you and you all go over the cliff.

    49. Re:Must past this test by dbet · · Score: 4, Funny

      Why would a self driving car ever drive off a cliff?

      I don't know, maybe life wasn't what it expected it to be?

    50. Re:Must past this test by Anonymous Coward · · Score: 0

      This doesn't seem that far fetched to me. You have to remember that most roads don't have a sheer cliff to one side. The car knows that sometimes it perceives the world wrong - just like humans sometimes do. So you have a common object that is known-harmful (the oncoming car) and a rare object that is harmful too (the sheer cliff), but only if it is actually there. So the car has to weigh not only the relative harm of the two options, it also has to weight the probability that it is wrong about what is going on. It is more likely to be wrong about something rare (the cliff) than something common (the car). If it always chooses to crash into an oncoming car if it SUSPECTS that it will otherwise drive off a cliff, this could lead it to choose to crash into oncoming cars even in cases where actually driving off the side of the road was perfectly save. So I don't think it is far fetched that an autonomous car would, very rarely, drive off a cliff to avoid an oncoming car even in the absence of any software bugs or serious sensor malfunctions.

      Of course, as you point out, that's not the point at all. Autonomous cars won't be perfect, but they don't need to be - they just need to be better than the average human. If they are better than the best humans, which eventually they will be, then not using them becomes irresponsible. Humans drive off cliffs too and if they do it more than autonomous cars then the cars are better in that respect. It's an open question how these things actually compare to human drivers TODAY, even if the numbers look OK so far.

    51. Re:Must past this test by Cro+Magnon · · Score: 1

      If self-driving cars *NEVER* swerve, that's a problem right there. If there is a big board in the road, and no traffic beside you, swerving is the right move. The reason braking is usually better is because there often IS traffic in the next lane, and the human might not have time or presence of mind to look before swerving in a tense situation.

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    52. Re:Must past this test by gorzek · · Score: 2

      Quite true. An autocar can't get distracted. It also has much more sensory data available than a human does, and it can gauge and react to that data inordinately faster than a human could.

    53. Re:Must past this test by Anonymous Coward · · Score: 1

      you ever want to turn off the automated driver?

      How about driving is fun and getting carried is not.

      A car is not a toy. If you drive for fun you are a hazard to those around you and you should probably not have a drivers license.

    54. Re:Must past this test by Theaetetus · · Score: 2

      So the self-driving wonder swerves right to avoid the other car and zooms off the cliff. A human driver would recognize that hitting the other car in this instance is the safer solution then to go careening off the steep cliff.

      Someone has never, ever taken an AI class. Or even an algorithm class dealing with risk. Here's how the calculation actually works (and by the way, that approach is about 20-30 years old). Every situation is assessed an impact value: driving into oncoming traffic, 0 (very bad); driving into the right ditch, 10; swerving into a legal lane, 50; etc. Every situation is given a set of possible actions, with each action having a probability of being completed successfully. The algorithm multiplies the outcome with the odds of achieving that outcome, and picks the highest value. You can set it up in different ways, but the idea is the same: multiply outcome severity with odds of achieving outcome, pick lowest combined risk/outcome. In your situation, driving off the cliff (which is assumed to be very bad, since the car can see a very steep drop-off with no bottom) is going to have a much worse outcome than hitting the car in front of it. Hitting the car in front of it is guaranteed, but so is driving off the cliff. As a result, the algorithm will make the automated car hit the car in front of it, rather than drive off the cliff.

      Not to mention that cars don't sleep, always behave optimally (according to the algorithms in place), and have no blind spots.

      Although I agree with your analysis, the question itself is flawed... It presumes that the self driving car is in a situation where (i) there's a truck immediately ahead, (ii) a truck immediately behind with failing brakes, and (iii) a motorcycle in the next lane (the question doesn't actually specify whether the motorcycle is pacing the car and traveling in the same direction or oncoming, but it's mostly irrelevant*). In order to face the dilemma of (a) crash off the cliff, (b) get smooshed between the trucks, and (c) kill the motorcyclist, the car has to be tailgating the truck and being tailgated by the other truck. Why did the AI let that happen? Why was it tailgating in the first place such that it can't speed up any to avoid the truck behind it, allowing it to pass the motorcycle and change lanes? Why was it content to be tailgated by the other truck without slowing down for safety reasons?

      The question sets up a situation in which the AI has already failed, and then asks whether the AI will fail. That seems misleading at best, and circular at worst.

      *only mostly: if the motorcycle is in an oncoming lane, then the car will be able to change lanes in a few seconds anyway. And if it's not tailgating the front truck, which it shouldn't be, then the car will have room to speed up to avoid the rear truck while waiting those few seconds.

    55. Re:Must past this test by Anonymous Coward · · Score: 1

      Thanks for saying this, it sums up my concerns exactly. This technology is too new and too unproven to actually put on the road with real people in cars and pedestrians. I can't imagine the worth of this technology being more valuable than even one single human life being taken by it.

      As a programmer, I know software is far from perfect. Bugs happen. I can't imagine the kind of certification and testing a program like this would need to be approved; you'd think it would be on par with military aircraft. But there is none.

      So the entire program works on the premise that Google's programmers aren't making any mistakes, catching their bugs, and have a testing methodology that works. Well, that might be the case. Or it it might not. Google certainly makes a lot of mistakes and many of its applications are buggy and remain buggy. Can the company that allows that, also let such bad habits trickle into the self-driving car division? Maybe.

      It's way too early to put self driving cars on the road.

      As a side note, if we want to reduce fatalities on the road we really should do more to stop drunk driving, drugged driving, and phones/texting while driving. This feels like a case where we are letting technology let us hold onto our vices because we are shifting the burden of responsibility onto the car.

      I mean sure, a self-driving car would let somebody who was totally shitfaced get home safely, but as a society do we really want people drinking that much? I'd rather see money go into program to help save alcoholics and keep the public aware about the dangers of drinking. You only have to watch 5 minutes of Hulu to see 100000 booze ads, after all. Same for drugged driving and people texting while driving. Technology shouldn't legitimize this behavior because it takes control of the wheel.

    56. Re:Must past this test by lister+king+of+smeg · · Score: 1

      I would be fine with an passive connection to other computing system, a stream of traffic data from other vehicles could be useful only as long as it is viewed by the cars guidance system as somewhat untrustworthy and firewalled. also it can serve up all of the data it likes to other components of the car so as you speed up the radio/sound system can raise the volume, or gas/battery (for electric vehicles) usage could be useful for telling saying low and giving another internet connected computer data that can make suggestions of nearby refill/charge station and dinning facilities

      But definitely as un-hackable as possible for example it should not trust reading sent to it by other cars without verifying them with say onboard radar and other sensors. And under no circumstances allow it to fallow command sent to it by another system (not counting steering wheel)

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    57. Re:Must past this test by 0xdeadbeef · · Score: 1

      does what its programming tells it to do, avoid hitting other vehicles. So the self-driving wonder swerves right to avoid the other car and zooms off the cliff.... I am amazed at how delusional governments are into so quickly allowing this technology on the roads... That just is irresponsible

      This is why direct democracy is a bad idea. Very often the government does in fact know better than morons with dumb opinions, and in this case, opinions about what constitutes safe driving that actually make him the very threat he accuses self-driving cars of being.

    58. Re:Must past this test by theRunicBard · · Score: 1

      What you're saying can be misinterpreted as "Because this hasn't reached perfection, it should not be allowed to progress." That's not true at all. Google's cars are already a shit-ton better than probably 30% (if not far more) of all drivers on US roads today. YES they should be allowed to progress. Yes there will be issues, but that's ok, because it's still better than what we have now, and it looks like people are being responsible. The quote even states that drivers are still legally required in the cars. How then, will that 100 car pile up you speak of result in the manufacturer being sued out of existence? It will be sad, true, and someone will be sued, but the manufacturer will just say "There were as many drivers in the car then as ever in the past. The driver could have taken over and didn't."

    59. Re:Must past this test by Anonymous Coward · · Score: 1

      If I was 'driving' the car and came accross a steep drop - I would take control.

      As someone mentioned below .. if your "driverless car" experience is to sit there waiting to take control of the device when you sense that it is about to get into trouble, then that is going to be a stressful and shitty experience. You might as well have been driving yourself all of the time.

      That is the situation everyone in your passenger seat is in.

    60. Re:Must past this test by gorzek · · Score: 1

      Yeah. I think the most utility of self-driving cars will come from daily commutes and long trips that consist primarily of interstate/highway driving. I would certainly like that.

    61. Re:Must past this test by mellon · · Score: 5, Insightful

      I think that the way it will play out is that as self-driving cars become a real and viable option, the penalties for bad driving will go up—drive drunk once, and you lose your license permanently, because why not—you can just use a self-driving car. Driver's tests will get harder, because why not—if you fail, you can just use a self-driving car. It will start with really egregious behavior, because voters won't feel threatened by it in sufficient numbers to cause a problem. Over time, the standards for human drivers will go up; at some point driving your own car will be about as common as flying your own airplane. We'll also probably stop giving licenses or learners' permits to teenagers, because they don't have the vote, and their parents would prefer to avoid a teenage testosterone tragedy.

      Of course, a really spectacular failure on the part of a self-driving car could put that whole scenario off by a generation.

    62. Re:Must past this test by Anonymous Coward · · Score: 0

      Yep, it just happens to be probabilistic hacking; doesn't work on everyone, but works.

    63. Re:Must past this test by NatasRevol · · Score: 1

      Soooooo, if I am a horrible driver, I can do those things?

      Cool!

      --
      There are two types of people in the world: Those who crave closure
    64. Re:Must past this test by ceoyoyo · · Score: 1

      You're right. There is no modern computer that can turn its clock frequency down within a few orders of magnitude of your brain's.

    65. Re:Must past this test by lister+king+of+smeg · · Score: 1

      tell that to the pollitians that will use it as fuel for their next ellection by running on the paltform of "I will stop those horrible self driving cars from trying to murder you by driving off of clifs"

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    66. Re:Must past this test by Anonymous Coward · · Score: 0

      That's retarded. You've basically said that if you enjoy driving then you're automatically bad at it.

    67. Re:Must past this test by Anonymous Coward · · Score: 1

      Quite true. An autocar can't get distracted.

      Of course it can. Feed it too much conflicting data and watch it dump the problem in the driver's lap. Where the driver will then drop his coffee and cell phone because he thought he was in a 'driverless' car.

    68. Re:Must past this test by h4rr4r · · Score: 1

      Odds are cliffs do not move often and any automated car will have access to maps with topo data.

    69. Re:Must past this test by gorzek · · Score: 2

      All excellent points. We already have computer-assisted driving. Automatic traction control and stability systems have computers hooked up to your car, monitoring the vehicle's characteristics at all times. They adjust in real-time to keep the vehicle on the road, going in the direction you have it pointed. They can do this a lot more effectively than a human ever could.

      It's time people realized there are just things machines are better at than we are. It's not something that denigrates humans, it's just accepting reality.

    70. Re:Must past this test by Aqualung812 · · Score: 4, Interesting

      Did you mean this forecast to sound neutral, damming, or hopeful?

      I see it as hopeful. Driving a car on a public road isn't a right. If you want to drive with manual control, do it on a road you paid for yourself.

      --
      Grammer Nazis - I mod you "troll" unless you actually add something on-topic. Yes, I know I have mispellings in my sig.
    71. Re:Must past this test by Anonymous Coward · · Score: 0

      "It's not a question of the algorithm. It's the question an algorithm."

    72. Re:Must past this test by h4rr4r · · Score: 1

      Because unless you drive fast enough for it to be challenging, where is the fun in it?

      It is only fun when it is dangerous.

    73. Re:Must past this test by NeutronCowboy · · Score: 1

      yep - the hard part is not the decision making. That's easy. The hard part is figuring out what it is that you're looking at. That's why computers do so well at limited scope problems like chess and checkers, and why they're still pretty dumb at understanding what they're interacting with.

      --
      Those who can, do. Those who can't, sue.
    74. Re:Must past this test by Anonymous Coward · · Score: 1

      Wow, not even close. The robo-vehicle can think a thousand times faster then a human driver, the nano-second the driver cuts the vehicle off, the robo-car will be able to break well before there is an issue, not SWERVE off the road, that isn't protocol. Even if said human driver rammed the driverless vehicle, the computer will control the situation 1000x better then someone freaked out they just got hit and over-correcting slamming on breaks. Not only this, but all the other computers around this scenario will be compensating and reacting within milisecond, instead of the 3-5 second delay the human break takes to even PROCESS the information. Basically you wrote a book attacking these vehicles, yet have done no research as to how they process information and react to certain situations. 100 car pile up? Really? Did you even think about that one before you wrote it?

      The more likely scenario is a couple of deaths of year from software issues. Even if ONE cars computer blows up or screws up entirely, the rest will be working properly, and compensating for said car's incident. To have 100 computers all shut down simultaneously, together, any thinking person knows has almost no chance of ever happening, ever. We already have 30-100 car huge fatality accidents now, so your point is completely invalid. On top of the MASSIVE reduction of drunk driving incidents, you have lambasted the issue overwhelmingly. The government you slammed for having these vehicles, has been having cross-country races with driverless vehicles, with the ONLY problems coming from HUMAN drivers ramming into their vehicles. Not once has one of these vehicles been in, much less caused a giant pile up, and they were in BETA testing! http://en.wikipedia.org/wiki/Autonomous_car do some reading before slamming something you obviously know nothing about, thanks.

    75. Re:Must past this test by Anonymous Coward · · Score: 0
    76. Re:Must past this test by h4rr4r · · Score: 1

      Your passenger seat has peddles and a wheel?

      In my car they are often on the phone or playing with their smart devices. They would not be ready to take over, nor does that side of the car have the needed controls.

    77. Re:Must past this test by thesandtiger · · Score: 1

      The problem is that the self-driving car wouldn't get into that situation in the first place, and only an extremely shitty human driver would, too.

      Any decent driver would anticipate that the other car might have to cut in early under conditions like that and would slow down so that if the other driver did need to cut in early they could.

      If you are driving in those conditions and you are taken by surprise by someone 1) coming up fast behind you and 2) going into the left lane to pass you and 3) cutting you off because of oncoming traffic you are simply put a spectacularly shitty driver and should not have a license in the first place. Nor should the other idiot, but in the scenario you describe it's a failure on the part of both drivers if there's a collision, let alone going off a cliff.

      --
      Since I can't tell them apart, I treat all ACs as the same person.
    78. Re:Must past this test by NeutronCowboy · · Score: 2

      And I forgot my drooling-from-the-mouth-fanboy/shill check list:
      * brand new account
      * posts a long post the minute the story goes live, despite the user not being a subscriber
      * subtle or over anti-Google bent in post

      sounds to me like there is some massive lobbying going on to short-cut the necessary amount of time to test auto-driven cars under all senarios, not just ones in controlled and predictable setups like we have seen.

      Ah, here it is. Google is paying off the government in order to kill us more quickly! Quick, bring out the pitch forks!

      Go away.

      --
      Those who can, do. Those who can't, sue.
    79. Re:Must past this test by CannonballHead · · Score: 1

      The computer will always drive the same, humans are not the reliable.

      Oh? Computers, which run humanly written software, never have problems, bugs, glitches, etc.?

      That said ... I do agree that a computer would be way more consistent than a human. But I would hesitate to say "always."

    80. Re:Must past this test by Anonymous Coward · · Score: 0

      The original post says there must be a licensed driver behind the wheel at all times ready to take over. They are also not yet allowing them to be sold to consumers. That is a long way from allowing droves of them to roam free on our roads. That really will be a long time coming; for instance just a fresh snowfall renders the car unable to detect driving lanes. I see this as the first step to allowing a few trained drivers to gather the kind of long-term, real world experience needed to slowly approach the dreams of truly autonomous vehicles.

    81. Re:Must past this test by Alotau · · Score: 4, Funny

      "upset at my wife" AND "in a hurry to get home"????

      This just proves how unreasonable human drivers can be.

    82. Re:Must past this test by Anonymous Coward · · Score: 0

      "I am amazed at how delusional governments are into so quickly allowing this technology on the roads"

      Dude, it's governor moonbeam. Probably the most irresponsible politician in the history of the state. He single-handedly threw CA's budget under the bus for the last 30 years by selling out to public unions in the early 1980s, signing onto ridiculous pension systems we have today. I'm convinced Jerry Brown would kill babies if a lobbyist could make him believe it was progress and donated $100k to the democratic party of California.

    83. Re:Must past this test by metrometro · · Score: 1

      Designing a specific test doesn't prove anything. Want a specific outcome in test scenario? Relatively easy. Testing real products against the actual environment, in which many accidents are unavoidable even with ideal driver input and comparing the failure rate and quality of those failures. It's not all quantitative either: is this a car accident, or a terrifyingly creepy car accident that makes people afraid of the cars?

      If there's going to be a company that blends math and design like this to create something life-or-death important, I'd rather have Google do it, than say, GM. I'd also want regulators to create clear rules for how results will be evaluated by the public. Which it sounds like we're doing. Getting this regulatory stuff right now is critical, because there aren't bazillions of dollars at stake yet and rational decisions have a chance of becoming law.

       

    84. Re:Must past this test by 0123456 · · Score: 4, Funny

      Odds are cliffs do not move often and any automated car will have access to maps with topo data.

      Given the last directions I got from Google Maps concluded with 'now drive through the barrier at the side of the highway and fall forty feet into the parking lot of the hotel below you', that does not give me warm fuzzies.

    85. Re:Must past this test by thesandtiger · · Score: 1

      Human brains get hacked all the time in ways that allow for traffic accidents:

      Medication, alcohol, fatigue, hunger, illness, self-delusion, optical illusions, distraction, emotion, etc.

      People get road hypnosis all the time when driving or are tricked into not noticing things that a computer with decent sensors absolutely would notice.

      And your reflexes are absolute shit; I don't care if you are at the peak of human reaction times, what you see as an instant reaction would, to any modern computer feel like an incredibly boring multiple month wait as your meat responds.

      --
      Since I can't tell them apart, I treat all ACs as the same person.
    86. Re:Must past this test by fastgriz · · Score: 1

      In your strawman scenario, the human driver is likely to not even be aware that she needs to make a decision because she's too busy texting while putting on makeup and blowing on a latte to cool it down.

    87. Re:Must past this test by gorzek · · Score: 4, Interesting

      That brings up another thing autocars will be better about than humans. Individual humans can learn from their mistakes, but that knowledge is not directly transferable to other humans. Any mistake a self-driving car makes, however, can have its solution incorporated into all self-driving cars (or at least all the ones of that model.) So, lots and lots of testing should ultimately give us very safe and effective cars.

    88. Re:Must past this test by flimflammer · · Score: 3, Insightful

      Are you an objective source for deciding whether or not you're a better driver than the machine?

    89. Re:Must past this test by 0xdeadbeef · · Score: 2

      The carpool lane was fortunately empty

      So you consider luck an algorithmic input?

      If the highway were non-divided you could have sent another car into oncoming traffic. And if not, into the wall, likely flipping it. You're rationalizing a mistake that you made as if it were optimal.

      If you don't have time to look, you don't change lanes, period. Unless it is a pedestrian in the road, you're more likely to kill someone swerving than by braking and possibly hitting the thing in front of you.

      Auto-driving cars, even with their greater awareness of other cars, still won't swerve like you believe they would, simply because they're more likely to lose control, and because the source of the danger is just as likely to keep going to the next lane.

    90. Re:Must past this test by FatAlb3rt · · Score: 1

      I think what we'll see is that Google cars will be uploading both human and computer-driven data to a service. This service will continually fine-tune the maps - so it can identify that there has been a stationary object in the same place the last 6 times (it's probably a mailbox, street sign, or light pole). So streets that weren't mapped before, or maybe didn't have enough data to be considered computer-safe, could become so rather quickly.

      If you're into statistics at least a little bit and have an interest, I strongly suggest Udacity's course on self-driving cars. It gives a lot of insight into how the car works.

    91. Re:Must past this test by superdave80 · · Score: 1

      The faster he gets home, the sooner he can smack her I guess?

    92. Re:Must past this test by Anonymous Coward · · Score: 0

      You are a small-minded idiot who allows fear to override whatever intelligence you may have.

      Now that we have that out of the way let's move on to what matters more.

      The self-driving car might not be ideal for all circumstances, but for some circumstances it will be
      a genuine boon. Perhaps you have never driven on I-70 across Kansas ? That road is very straight
      and involves little elevation change, and it would be absolutely wonderful for a solo driver to have the
      option to sleep a bit while the car took car of the task of driving.

      In any case, the world WILL embrace this technology whether you like it or not. It's called
      progress, and it has always driven over fools like you as a steamroller drives over a section of
      road. Get used to it and adapt, or get the fuck out of the way.

    93. Re:Must past this test by FatAlb3rt · · Score: 1

      You should take into account that a self-driving car is aware that the other car is moving toward your car before you realize it. And it is much more likely to take appropriate evasive action (perhaps just mild braking) much quicker than you can.

    94. Re:Must past this test by Anonymous Coward · · Score: 0

      Slashdot needs to put commas in those numbers. I keep seeing two million as two hundred thousand.

    95. Re:Must past this test by Anonymous Coward · · Score: 0

      So every new poster's first comment is a troll?

      Grandparent post is wrong, certainly. But it isn't a a troll.

    96. Re:Must past this test by eth1 · · Score: 1

      Honestly, I'd probably be fine with them once they're statistically safer than human drivers.

      What I would be most worried about is that they're too predictable - once the rules they use are known, they can be abused. "Self-driving car baiting" would be my fear. Some ass gets road-ragey for some reason, and decides it would be funny if he takes advantage of some known method of getting another car to stand on the brakes in 70 MPH traffic, or something.

    97. Re:Must past this test by Anonymous Coward · · Score: 0

      Why would a self driving car ever drive off a cliff?

      I don't know, maybe life wasn't what it expected it to be?

      The above is easily the most hilarious comment on Slashdot in many months.

      If it is not modded up further it is because the readership these days is a bunch of
      idiots like "dadinportland" and other such empty-headed fools.

    98. Re:Must past this test by drerwk · · Score: 4, Insightful

      The system just needs a rapid manual override and a little common sense from the driver.

      See the results of the http://en.wikipedia.org/wiki/Air_France_Flight_447 AF447 flight for the odds of this working. As a one time private pilot I am totally baffled as to how a professional pilot could hold a plane in a stall from 35,000 ft to the ground. I think there were several issues including human factors in the design of the interfaces; but I really think that these guys got used to being along for the ride and it was not conceivable to them that the plane had decided to stop flying itself.

      After a week of having an auto-car drive me to work everyday I can not imagine I'd be ready in 1/2 second to suddenly take over for the computer and expect a good result.

    99. Re:Must past this test by Zordak · · Score: 5, Funny

      Happens all the time. Your forgot to turn off the "Professsional Stuntman" option. For some reason they have that box checked by default. You might also want to double check the settings for "Knight-Rider Style Turbo Boost" and "Assume I Have Access to Airwolf."

      --

      Today's Sesame Street was brought to you by the number e.
    100. Re:Must past this test by Anonymous Coward · · Score: 0

      Well your first problem was that it was Apple Maps, and your second problem was that you were looking at the icon.

    101. Re:Must past this test by Anonymous Coward · · Score: 1

      Because unless you drive fast enough for it to be challenging, where is the fun in it?

      It is only fun when it is dangerous.

      A certain indicator of a weak intellect is the inability to imagine
      circumstances which might fall outside your own personal experience.

      That's right, harrar, I am saying you are STUPID.

      I enjoy driving as perfectly as possible while obeying all traffic laws.
      This means matching revs with each double-clutched downshift, hitting the apex
      in a curve precisely, and other advanced driving techniques with which you
      may not be familiar.

      Bottom line : just because things do not conform to the worldview displayed on the miserable
      low-res screen in your tiny little brain does not mean you are correct.

    102. Re:Must past this test by drerwk · · Score: 1

      Braking has two additional advantages; 1) removing energy from an impending collision, and 2) increasing the time in which one has to react to the next issue, because you will feel pretty dumb if you let yourself get boxed in to a collision after missing the first obstacle. Of course one should not brake so hard as to get hit from behind.

    103. Re:Must past this test by ryanmc1 · · Score: 1

      Whoever said home is where his wife is? He probably left her at the store and is hurrying home to pack his stuff before she catches up.

    104. Re:Must past this test by Zordak · · Score: 2

      If I have a hammer I have to use it for every single task I have, otherwise what's the point of owning a hammer?

      I'm a little confused. Can you restate that with a car analogy?

      --

      Today's Sesame Street was brought to you by the number e.
    105. Re:Must past this test by geddo · · Score: 2

      I think the point of the self-driving car is for congested roads like freeways. I am not too keen on trusting software to drive my car even on a freeway, but I could change my mind if it meant a faster commute. I could see where a system is initiated prior to entering the freeway, auto-driven cars get a lane like the HOV and send out a signal to receivers on the road identifying it as being auto-driven and permitted in the lane, once the vehicle leaves the freeway the driver could switch back to manual mode. Likely this would expand to highways as well but I think the return is diminished when you go from multi-lane freeways to one or two lane roads, IMO its primarily the people who cannot merge at speed that cause the freeway backups. I cannot imagine ever trusting a software program to drive me through canyons and cliffs, besides driving canyons and cliffs are fun. You want to test your accident avoidance software and impress me, log thousands of miles on our Cali freeways without an accident driving a motorcycle, then I'll be impressed.

    106. Re:Must past this test by CanHasDIY · · Score: 1

      1) Reaction time to a specific, anticipated event, you're likely right. But there's one crucial element to navigation that humans can do that computers currently can't - make arbitrary, seemingly illogical decisions in the heat of the moment. Take GP's hypothetical about the cliffside road - most humans, thanks to self-preservation instinct, will choose to rear-end the other car rather than drive off the ledge; what would a computer that is programmed to "avoid contact with other cars at all costs" do in that situation? Hyperbole aside, there's no way of knowing until we put one in that real world situation.

      2) I'll give you that one, I suck at distance estimation, although I wouldn't go so far as to imply that machine calculations are infallible, either. They are programmed by humans, and subject to elemental factors, after all.

      3) If vision is the only sense you use when driving, you shouldn't be driving.

      4) an auto-car may maintain speed better, assuming that it is receiving accurate information from its sensors, and interpreting said information correctly. What happens when the auto-car's front facing range finder gets just enough mud on it to make the car think there's an object .2mm from the front end of the vehicle? Will it refuse to move forward, or perhaps start reversing furiously to avoid the "obstacle" it's certain it's about to smash into?

      5) A hacker doesn't need to gain access to the actual auto-driving mechanism to cause problems (though, considering the fact these things use Wifi and other forms of RF communication betwix themselves, it wouldn't be a stretch to think they could be compromised), all the perp would need to do is provide enough false or junk data to confuse the auto-driving algorithm into abnormal behavior. This can be done with readily available, off the shelf electronics.

      Speaking of which... how do you suppose an auto-car would handle a construction blasting zone, in which drivers are legally required to shut down all radios?

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    107. Re:Must past this test by CanHasDIY · · Score: 0

      1) says some anonymous person that doesn't know me from Adam. Anyone capable of cogent thought will take your statement with a heaping serving of NaCl.

      2) Advertisers are not scientific researchers. They are morons, just like the suckers who fall for their BS.

      Any more ad hominems or non sequiturs you care to throw out there?

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    108. Re:Must past this test by CosaNostra+Pizza+Inc · · Score: 1

      But for every-day driving, does it need to be better than you? Instead of wasting an hour or more every day with your attention on the road, wouldn't you rather be reading, talking on the phone, texting, preparing for your work presentation/speech, doing your homework, etc?

    109. Re:Must past this test by gorzek · · Score: 1

      If self-driving cars are found to be safer than human drivers (and I think they will be), then your insurance company might actually give you a discount. You would still pay for insurance like normal, but it would be based on how well your autocar performs during accidents.

      If there is found to be a flaw in the car itself (such a bug that makes it speed up uncontrollably, plowing into pedestrians or something) then obviously that is a recall/lawsuit situation. Manufacturers would be just as liable for that as they are for any other demonstrable defects in their vehicles.

    110. Re:Must past this test by mooingyak · · Score: 1

      "upset at my wife" AND "in a hurry to get home"????

      This just proves how unreasonable human drivers can be.

      She could be in the car with him.

      --
      William of Ockham had no beard. The most likely explanation is that it was chewed off by squirrels every morning.
    111. Re:Must past this test by Anonymous Coward · · Score: 0

      If you drive for fun you are a hazard to those around you and you should probably not have a drivers license.

      Is driving really such a hassle to you that you're unable to get any enjoyment out of it? Unless I'm stuck in traffic, I enjoy every second of it. Cruising along the highway with music filling the interior, a cool breeze running along my chin, no one to bother me except the occasional silly driver that you simply avoid. There's nothing I find more relaxing other than deep sleep.

      I would ask what you're smoking, but in your case I think some relaxation ought to be good.

    112. Re:Must past this test by geoskd · · Score: 1

      Happens all the time. Your forgot to turn off the "Professsional Stuntman" option. For some reason they have that box checked by default. You might also want to double check the settings for "Knight-Rider Style Turbo Boost" and "Assume I Have Access to Airwolf."

      You forgot to turn off the "Can swim 3000 miles without drowning option". Looks like your new parking space is significantly below sea level...

      --
      I wish I had a good sig, but all the good ones are copyrighted
    113. Re:Must past this test by Anonymous Coward · · Score: 0

      Well, leaving out the car in the other lane WAS a critical piece of information.

    114. Re:Must past this test by CanHasDIY · · Score: 1

      The quote even states that drivers are still legally required in the cars

      Um... if the car drives itself, wouldn't that technically make the occupants passengers?

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    115. Re:Must past this test by geoskd · · Score: 1

      And sometimes it's just more fun to drive the car yourself.

      I'll bet its pretty fun to drive while under the influence of your favorite drug too, that doesn't mean its a good idea, or should be tolerated.

      If it can be demonstrated that autopilots are significantly safer than human drivers, then manual drivers are by definition endangering the public welfare for their own amusement. Sounds like the functional definition of a sociopath to me...

      -=Geoskd

      --
      I wish I had a good sig, but all the good ones are copyrighted
    116. Re:Must past this test by Anonymous Coward · · Score: 0

      ... and sometimes it gives you wonderful suggestions. http://xkcd.com/461/

    117. Re:Must past this test by Anonymous Coward · · Score: 0

      Any decent driver would anticipate that the other car might have to cut in early under conditions like that and would slow down so that if the other driver did need to cut in early they could.

      ^ Could be rephrased as

      Any decent driver would be able to read the minds of every other driver on the road, who might have to cut in early under conditions like that and would slow down so that if the other driver did need to cut in early they could.

      You, sir, are a jackass.

    118. Re:Must past this test by camperdave · · Score: 1

      There is another option the autocar can take: apply the brakes and slow down. That will allow the passing car enough room to get into the lane. Chances are high that the computer would have looked at this scenario already, and have slowed down in anticipation.

      --
      When our name is on the back of your car, we're behind you all the way!
    119. Re:Must past this test by HideyoshiJP · · Score: 1

      Not really. You should try driving a vehicle with a manual transmission. The most mundane of drives can feel exciting - without speeding.

    120. Re:Must past this test by Krishnoid · · Score: 1

      the penalties for bad driving will go up—drive drunk once, and you lose your license permanently, because why not—you can just use a self-driving car.

      One example of when you take those penalties to their inevitable conclusion.

    121. Re:Must past this test by geoskd · · Score: 2

      'Why did you turn off the computer when you know it is proven to be safer?'

      "Because my brain operates at a frequency modern computers cannot even begin to match, and it cannot be hacked."

      And yet somehow, in spite of that, you have just demonstrated exactly why You should not be allowed to operate a lethal weapon on our streets. When it comes to objective evaluation of the situation, you are fail... Paranoia != Righteousness

      -=Gsoekd

      --
      I wish I had a good sig, but all the good ones are copyrighted
    122. Re:Must past this test by mooingyak · · Score: 1

      It seems that right now it only recognizes the possibility of crossing the Pacific. I asked for NY --> London directions and it sent me West across the US, across the Pacific to Japan, then the same instructions from yours into China, and then through Asia and Europe to the UK. Crossing the Atlantic would be shorter (didn't it used to suggest that?)

      --
      William of Ockham had no beard. The most likely explanation is that it was chewed off by squirrels every morning.
    123. Re:Must past this test by HideyoshiJP · · Score: 1

      5) I'd expect an auto-driver system to be seperate from any other computing devices in the car, and connected to the internet or any other vector for hacking. I'd expect them to be as immune to hacking as an auto-pilot system in a plane.

      While you'd expect it to be separate, the track record for the security of vehicle electronics is not good at the moment. Unless there's some regulation on the matter (be it government mandated or industry self-regulated), the bean counters may not see the benefit of doing so until Consumer Reports does some exposé on the matter and everybody starts to panic over it.

    124. Re:Must past this test by Rhacman · · Score: 1

      My argument isn't that we don't have the technology, only that machines can't be imbued with accountability for their actions. The ONLY scenario where I can see an automatic driving system useful in is if the system can detect that the driver has fallen unconsious. There is NO replacement for a driver who maintains full situational awareness of the driving environment. That means eyes on the road, checking mirrors, and ready to take control of the vehicle instantaneously should an unexpected situation arise. No texting, no alcohol, no entertaining the kids. If you are already doing all of this, then just put your foot on the pedal and your hands on the steering wheel and just drive! You can't delegate accountability to a machine. If they want to design a system that can bring the car safely to the side of the road if the driver becomes unresponsive then fine, but don't tempt people into thinking they can have a robo-chauffeur drive them home from the bar or take little Billy to school by himself.

      --
      Account -> Discussions -> Disable Sigs
    125. Re:Must past this test by Idbar · · Score: 1

      Let me also clarify something that some people take as a given:

      6) I also expect the auto-driver concentrated on its job instead of reading a book or checking the news off its tablet or talking on the phone or texting or staring the hot guy/chick.

    126. Re:Must past this test by rwv · · Score: 1

      Will an auto-car be able to distinguish hazards like pot holes, road kill, live animals (and people) in the road, swerving drivers, construction zones, and any situation where crossing the double yellow line is necessary?

      Will an auto-car be able to be directly safely by a police officer controlling a traffic situation? Will an auto-car be able to navigate toll booth plazas? Will an auto-car be able to navigate roads when traffic signals aren't working because of a power outage?

      Here's my test for auto-car road-worthiness... Have the car find parking in a crowded suburban mall parking lot on a weekend in December. If it can safely enter the lot, find a spot, and exit the lot ten times then it gets approval for everyday life.

    127. Re:Must past this test by HyperQuantum · · Score: 1

      But someone who is in a hurry to get home is likely to switch off the auto-pilot and drive manually, because it "drives too slowly".

      I agree with you though, but we need to keep in mind that simply introducing self-driving cars is not a panacea.

      --
      I am not really here right now.
    128. Re:Must past this test by HyperQuantum · · Score: 1

      Question: will the programmers of the car's driving software be required to have such a (high standard) driving license to be allowed to work on the software?

      --
      I am not really here right now.
    129. Re:Must past this test by Patch86 · · Score: 1

      I'm 99% sure that someone from the Google Maps Team is a Slashdot regular- it looks fixed to me now.

      Not the first time someone's bitched about a Google Maps bug on here only to see it immediately fixed (happened to me once).

    130. Re:Must past this test by SolitaryMan · · Score: 1

      "Avoid hitting all cars" is just plain bad directive. For example, I, a human driver, don't care too much about hitting other cars as long as it is not my fault. Of course, I try not to hit stupid assholes not using their turn signals and all, but I don't try too hard either.

      --
      May Peace Prevail On Earth
    131. Re:Must past this test by tnk1 · · Score: 1

      Why would a self driving car ever drive off a cliff?

      It shouldn't. On the other hand, it may not be aware of the fact there is a cliff there, as opposed to a less precipitous drop-off. There are ways to make sure that doesn't happen, but it is definitely something that should be considered. Presumably, Google already has, but it can't hurt to ask the question.

      As another situation, I am very interested in how these cars handle icy driving conditions. I've been hit from behind by someone who could not safely stop their car on an icy surface, and it would be nice if the cars could prevent those sorts of problems caused by weather conditions.

      I'm all in favor of automatically driven cars. I just hope that they eventually use the opportunity to let those cars drive much faster under automatic control than human operated ones. Either way, it should be better no matter what if they just make sure the cars all follow a set cruising speed and you don't have to worry about someone backing up traffic in the passing lane, and someone else tailgating you.

    132. Re:Must past this test by stanlyb · · Score: 1

      I see only way for the human brain to manage this situation: Having a bird-eye camera.

    133. Re:Must past this test by SolitaryMan · · Score: 3, Insightful

      I wonder how insurance companies are going to handle this. My self-driving car hit your self driving car. Who's going to pay? Yeah, my car is at fault, but I wasn't at the wheel and I don't even have a license. What then? What if a collision is due to a bug in software?

      I'm afraid that legal obstacles this project faces are more serious than technical.

      --
      May Peace Prevail On Earth
    134. Re:Must past this test by Sir_Eptishous · · Score: 1, Informative

      'Why did you turn off the computer when you know it is proven to be safer?'

      I was using the Force, what else?

      --
      We play the game with the bravery of being out of range
    135. Re:Must past this test by mooingyak · · Score: 1

      I still see the westbound route from NY to London...

      --
      William of Ockham had no beard. The most likely explanation is that it was chewed off by squirrels every morning.
    136. Re:Must past this test by clyde_cadiddlehopper · · Score: 1

      3. You get bored, distracted, and fall asleep behind the wheel ... especially on long straight highways. 4. You are limited in how many hours per day you can drive. 5. You are expensive. This is why the premier application (note I did not use the term "killer app") for self-driving vehicles will be in long haul trucks.

      --
      Obi-Wan: "I felt a great disturbance in the Force, as if millions of voices suddenly cried out in terror and were sudden
    137. Re:Must past this test by SolitaryMan · · Score: 1

      It would be interesting to make you compete against Self-Driving car for a Driving License :)

      Like, you drive as usual, but in a car with all self-driving sensors and all and then expert compares decisions you made with decisions your car was suggesting at the same time. If you can't beat it, you don't get the license :)

      --
      May Peace Prevail On Earth
    138. Re:Must past this test by misterooga · · Score: 1

      Yes, please. Pics or it didn't happen.

      On a more serious note, I wonder if self driving cars can also communicate with each other. Of course, this could go either way: speed up to becoming a police nation, or it can alert cars ahead of an incoming dangerous driver.

    139. Re:Must past this test by BasilBrush · · Score: 1

      1) Reaction time to a specific, anticipated event, you're likely right. But there's one crucial element to navigation that humans can do that computers currently can't - make arbitrary, seemingly illogical decisions in the heat of the moment. Take GP's hypothetical about the cliffside road - most humans, thanks to self-preservation instinct, will choose to rear-end the other car rather than drive off the ledge; what would a computer that is programmed to "avoid contact with other cars at all costs" do in that situation? Hyperbole aside, there's no way of knowing until we put one in that real world situation.

      The concept that an auto-driver would be programmed to "avoid contact with other cars at all costs" and not have a concept of cliffs vs level verges is not a sensible one.

      3) If vision is the only sense you use when driving, you shouldn't be driving.

      Well lets see. Apart from vision we have hearing, sight, touch, smell, taste, temperature, pain, balance, acceleration. Some of these are of little use in driving, and some of them are compromised in human drivers (e.g. soundproofing in modern cars, and the playing of music largely dicards sound as a useable sense by the average driver. But of all the driver-useful ones, technology gives us better sensors then the human ones.

      What happens when the auto-car's front facing range finder gets just enough mud on it to make the car think there's an object .2mm from the front end of the vehicle?

      Not being a designer of such systems I can't answer that. But it is a pretty obvious hazard, and can be assumed to be dealt with in a reasonable way. I do know that there is the principle of fail-safe and redundancy, so I'd expect a car with the loss of one of it's sensors to still be able to function safely, but to alert the human to the fault, and possibly to drive to a safe parking place and refuse to continue on automatic until the fault has been rectified.

      5) A hacker doesn't need to gain access to the actual auto-driving mechanism to cause problems (though, considering the fact these things use Wifi and other forms of RF communication betwix themselves, it wouldn't be a stretch to think they could be compromised), all the perp would need to do is provide enough false or junk data to confuse the auto-driving algorithm into abnormal behavior. This can be done with readily available, off the shelf electronics.

      I'd expect information fed via radio from other auto-drivers to be treated like the information a good human driver receives from other car's turn signals. Useful advisory information for maintaining a smooth drive, but not to be relied upon as totally reliable. So a safety margin is still left, regardless of what the other car actually does. If the information is wrong, I'd expect it to result in an emergency stop, rather than a collision.

      When I say "I'd expect" what I mean is that unless and until such systems behave like this or better, I wouldn't expect them to pass the safety tests and be licensed for consumer use. And I mean that I believe these things are possible with today's technology.

      In other words I don't see anything that's been suggested here as showstoppers, but just statements of some already known parts of the problem-domain.

    140. Re:Must past this test by BLKMGK · · Score: 1

      I'm amazed that with so little insight into how this car reacts in any scenario, much less the straw man you've proposed, that you've been able to come to a conclusion as to it's safety or lack thereof. Please provide some citations as to why you think the car would turn and go off the cliff - you're imagination doesn't count. Where did this imaginary "avoid hitting all cars" rule come from exactly? If it were up to folks like yourself we'd still be riding horses!

      --
      Build it, Drive it, Improve it! Hybridz.org
    141. Re:Must past this test by LateArthurDent · · Score: 2

      The question is not is it better than some hypothetical human driver, but is it better than the drivers we have right now.

      No, the question is: is it better than me?

      If not, I don't want it driving my car.

      It is.

      You're not that great of a driver. Being human prevents you from being a better driver. You only have eyes in front of you, and you need to turn your head and look around, pay attention to mirrors, each time taking your attention away from where you are going for a fraction of a second. The computer can pay attention to 360 degree sensors 100% of the time. Once you detect the need to take immediate action, you need to move your leg to hit the brakes. For the computer controlling the car, the brakes are accessible immediately upon determining they need to be used.

      At peak performance, a well rested and attentive driver will still not be as good as a well programmed machine. You can argue the machine isn't well programmed, but they're already better than the average driver, and every single mistake that results in a crash would result in a software update to every other self-driving car that is now guaranteed to not make the same mistake. So they'll become better than the best driver in no time once they are in common usage.

    142. Re:Must past this test by Anonymous Coward · · Score: 0

      I think there is far too much disagreement with this post from people who haven't designed safety systems. It's all easy in simulation, it's when you know your code is pacemaking someone's heart, preventing traffic accidents, keeping planes in the sky, etc. that you start sweating. Fully self-driving cars will be like fusion power, which has been 50 years away for more 50 years now. All it takes is one single programming mistake, hardware failure, or unfortunately timed bird dropping on your lidar that results in death or injury, and it's delayed by 10 years. The comparison to the failure rate is not valid when people start dying: in this scenario humans are allowed to fail, automated safety systems are not. In addition, there WILL be bugs in the first generations of the products, and people will not read all the fine print and will start shaving, sleeping, drunk, etc. in cars that were really designed (and tested!) assuming an alert attentive human driver is always present and waiting to take over.

      My favorite example out of the thousands of tricky cases is the "multiple visible traffic lights" problem. There are many times where you can see multiple red and green lights at one time (for example a long street, or a more than 4 way intersection in cities). How do you choose?

      Let's say a traffic law is violated by a self-driving car, who pays the ticket and who gets the points on their licence?

    143. Re:Must past this test by BaronAaron · · Score: 1

      A good driver will notice a dangerous situation and take manual control well before a bad event happens. For example, traffic congestion, road repair, bad weather, etc. Humans are still better at anticipating unexpected situations.

      If you are in auto-drive and a bad event happens that requires a split second action, as others pointed out, it's probably statistically better to let the computer figure out the best course of action.

      As far as auto-drive failure. If the system went offline, it would most likely be during a non critical time where the driver can safely transition to manual mode. Unlike a plane, a car can move to the shoulder, stop, and wait for the driver to get his or her act together.

    144. Re:Must past this test by Anonymous Coward · · Score: 0

      If it has to be better than you to be worth it, that means that your own driving skills are at the minimum acceptable level. Don't you find that a little unsettling?

    145. Re:Must past this test by CanHasDIY · · Score: 1

      3. You get bored, distracted, and fall asleep behind the wheel ... especially on long straight highways. 4. You are limited in how many hours per day you can drive. 5. You are expensive. This is why the premier application (note I did not use the term "killer app") for self-driving vehicles will be in long haul trucks.

      Not that I agree or disagree, but something tells me the trucker unions will have a thing or two to say about that...

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    146. Re:Must past this test by BasilBrush · · Score: 1

      I only see that as an intermediate step, whilst people become comfortable with the concept of auto-drivers. Of course that may require a generational shift of the population, rather the changing the minds of he current population of drivers.

    147. Re:Must past this test by BLKMGK · · Score: 1

      I'm betting that like 99% of the idiots on the road - including myself - that you rank your driving skills FAR higher than they actually are... Your's is a pretty emotional reaction and while I'm not sure I'm thrilled about having a self driving car - I enjoy driving - I'm certainly open to the idea and willing to consider it. Considering the traffic situations in some areas having cars drive themselves and coordinate among themselves makes a great deal of sense. At this point you have NO idea how well this car drives or doesn't but are jumping up and down screaming NO. It's kind of amusing actually.

      --
      Build it, Drive it, Improve it! Hybridz.org
    148. Re:Must past this test by h4rr4r · · Score: 1

      I am saying you are boring.

      I am quite familiar with double clutching and hitting the apex, and toe and heel. All pointless at highway speeds.

      Bottom line: Boring person, you likes boring things, driving pointlessly at low speed.

    149. Re:Must past this test by BasilBrush · · Score: 1

      Will...

      Why not? All of these things are individually possible with today's technology.

    150. Re:Must past this test by h4rr4r · · Score: 1

      Why would they?
      Great reaction times and snap judgements are about the opposite of good programming.

    151. Re:Must past this test by jkflying · · Score: 2

      I second that a computer has a faster reaction time than you do. Sorry, man, you seem outnumbered on this one.

      --
      Help I am stuck in a signature factory!
    152. Re:Must past this test by h4rr4r · · Score: 1

      I see a joke some aspies are not getting.

      The swim 3000 miles thing is a joke, google did it for a laugh. Much like zooming too close on the moon at one point showed cheese.

    153. Re:Must past this test by Martin+Blank · · Score: 2

      I fly a small plane for fun. I'm certainly more dangerous up there than a military or airline pilot, and it's far less efficient.

      At some point, you have to let people do what they're going to do even if there's some measurable risk to the public. It's a fine balancing act sometimes, but it's also part of a free society.

      --
      You can never go home again... but I guess you can shop there.
    154. Re:Must past this test by Anonymous Coward · · Score: 0

      Please list your hobbies so we can point out how boring and pathetic of a human being you are. You're being the lowest of the low at the moment.

    155. Re:Must past this test by BLKMGK · · Score: 1

      Odds actually are that the car sensed someone passing and actually SLOWED DOWN like humans are supposed to do in order to assist the passing vehicle. In fact the computer might have even sensed the oncoming car in the other lane and slowed even more to assist the idiot driver. Why the straw-man concludes that the car will never choose to strike another 100% of the time or that it won't see the situation coming and avoid it better than a human escapes me. He didn't write the code but has an extreme knee jerk reaction and a pretty tenuous imagined situation to try and support his ideas. It's actually pretty funny... :-)

      --
      Build it, Drive it, Improve it! Hybridz.org
    156. Re:Must past this test by bbn · · Score: 1

      That is for the insurance companies to figure out. If they can't, they will pay half each. Or they will sue each other. You wont care because your insurance wont go up since it wasn't your fault either way.

      There is no problem here. It is not even any different from having a car today. What if your car has a fault that makes it accelerate unexpectedly and crash into someone (Toyota!)? The answer is your insurance will pay, then they will sue Toyota for the damages.

    157. Re:Must past this test by mooingyak · · Score: 1

      I see a joke some aspies are not getting.

      The swim 3000 miles thing is a joke, google did it for a laugh. Much like zooming too close on the moon at one point showed cheese.

      And the walking directions for Mordor.

      I get the joke, just wondering why it's telling me to sail and jet ski across the Pacific instead of the Atlantic for NY->London.

      --
      William of Ockham had no beard. The most likely explanation is that it was chewed off by squirrels every morning.
    158. Re:Must past this test by Anonymous Coward · · Score: 0

      Well, the California legalization bill includes total immunity of all criminal and civil liability attached to accidents that are caused by driverless cars.

      Since there is still a requirement for there to be a driver in the car, all of the liability rests with them. Google and the auto manufacturers are completely immune, at least in California.

    159. Re:Must past this test by Anonymous Coward · · Score: 0

      That's interesting, because I'd expect an auto driving system to be hooked up to a lot of potential attack vectors. Let's see...

      1. Traffic data feeds
      2. Realtime map update feed
      3. Car-to-car ad-hoc networking for improved sensor range and lane negotiation
      4. A government override to disable auto-driving in construction or accident zones
      5. The various communication channels these streams are delivered over (WiFi, 3G, WiMax, local mesh network)

    160. Re:Must past this test by Anonymous Coward · · Score: 0

      Software that is intended for mission critical use is highly reliable and as close to bug free as you could get. Medical and space equipment come to mind. You would hope the same level of care will be given to the software driving cars.

    161. Re:Must past this test by Martin+Blank · · Score: 1

      It wasn't luck that the carpool lane was empty. It was a fact of the time of day (or night in this case) that it was basically unused. I was aware of it as I try to maintain a position in lanes that allows me the room to maneuver if necessary without hitting something to the side. I'm uncomfortable driving when that's not an option and I'm going more than about 35MPH.

      Most losses of control happen from improper maneuvers, like swerving but not turning the car back toward the original direction of travel. There are a lot of cases where swerving is not the optimal choice but people do it anyway. I've seen a couple of accidents that could have been avoided if the person had swerved into the shoulder, and I've done the same thing not because I was forced to do so but because I deemed it prudent as traffic was packed too tightly for everyone to be sure of safely stopping. I got myself out of the way and gave other drivers behind me another car-length of stopping distance. An auto-driving car should take such factors into account as it's not always that you're going to hit someone else, but that someone may hit you.

      Most vehicles today have very good stability. That stability, when measured many times per second, can be used to make maneuvers that on their own are abnormal or difficult for a normal person but perfectly safe for a professional driver or a computer. They may not happen as often, or they may happen more often. Only time will tell. But they must have the ability to do so because braking on its own isn't always the best reaction.

      --
      You can never go home again... but I guess you can shop there.
    162. Re:Must past this test by jkflying · · Score: 1

      1) Reaction time to a specific, anticipated event, you're likely right. But there's one crucial element to navigation that humans can do that computers currently can't - make arbitrary, seemingly illogical decisions in the heat of the moment. Take GP's hypothetical about the cliffside road - most humans, thanks to self-preservation instinct, will choose to rear-end the other car rather than drive off the ledge; what would a computer that is programmed to "avoid contact with other cars at all costs" do in that situation? Hyperbole aside, there's no way of knowing until we put one in that real world situation.

      Do a quick search for 'fuzzy logic'. It sorts out all the issues that you're talking about here. This problem has already been solved.

      2) I'll give you that one, I suck at distance estimation, although I wouldn't go so far as to imply that machine calculations are infallible, either. They are programmed by humans, and subject to elemental factors, after all.

      If my radar system isn't working because of the rain/snow, trust me, you're eyes won't be working either.

      3) If vision is the only sense you use when driving, you shouldn't be driving.

      So you're using sound and touch as well. Impressive, especially considering that there is no way in hell computer systems can have sound and touch inputs. Oh wait...

      4) an auto-car may maintain speed better, assuming that it is receiving accurate information from its sensors, and interpreting said information correctly. What happens when the auto-car's front facing range finder gets just enough mud on it to make the car think there's an object .2mm from the front end of the vehicle? Will it refuse to move forward, or perhaps start reversing furiously to avoid the "obstacle" it's certain it's about to smash into?

      It will probably tell you to clean the mud off of the sensor, or use a wiper/spray to do it for you.

      5) A hacker doesn't need to gain access to the actual auto-driving mechanism to cause problems (though, considering the fact these things use Wifi and other forms of RF communication betwix themselves, it wouldn't be a stretch to think they could be compromised), all the perp would need to do is provide enough false or junk data to confuse the auto-driving algorithm into abnormal behavior. This can be done with readily available, off the shelf electronics.

      So, how do you propose to disrupt RADAR, LIDAR and camera based systems in a way that wouldn't affect human drivers?

      Speaking of which... how do you suppose an auto-car would handle a construction blasting zone, in which drivers are legally required to shut down all radios?

      By using the LIDAR and camera based systems, leaving out the RADAR and pre-downloading map data. Easy peasy. They can even leave GPS running.

      --
      Help I am stuck in a signature factory!
    163. Re:Must past this test by Cro+Magnon · · Score: 1

      But for every-day driving, does it need to be better than you? Instead of wasting an hour or more every day with your attention on the road, wouldn't you rather be reading, talking on the phone, texting, preparing for your work presentation/speech, doing your homework, etc?

      But, don't most drivers already do all that?

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    164. Re:Must past this test by Skynyrd · · Score: 1

      The question is not is it better than some hypothetical human driver, but is it better than the drivers we have right now.

      No, the question is: is it better than me?

      If not, I don't want it driving my car.

      Ahhh, but the point is not for it to drive my car, but for it to drive the car of all those other people who don't pay attention to their driving. I'm not interested in one myself, but I want them to be legal and popular. I can't imagine how overall it would be worse for the rest of us.

    165. Re:Must past this test by FatAlb3rt · · Score: 1

      You see people driving and texting now - do you think that the distraction level will stay the same? No, it will increase. Greatly. People will be playing on their phone, reading the paper, maybe even getting dressed on their morning commute. I think it's completely unrealistic to think that people will be there at the ready, for the opportunity to take over from the computer.

    166. Re:Must past this test by Ogive17 · · Score: 1

      If these cars are being programmed to swerve in order to avoid a collision, I do not want to ride in one.

      They should brake only, swerving tends to lead to worse accidents. I'm also placing my faith in the car that its' reaction time would be far superior to mine, so that braking only would avoid more collisions than I would.

      Now that I've thought about my 17 years behind the wheel, not once have I ever had to swerve to avoid a collision. My only accident was my fault when I pulled out in front of someone. I once hit a groundhog at night on a country road, a swerve would have put me into a ditch and still killed the groundhog. (didn't see groundhog until last second).

      --
      "Action without philosophy is a lethal weapon; philosophy without action is worthless."
    167. Re:Must past this test by LateArthurDent · · Score: 1

      Reaction time to a specific, anticipated event, you're likely right. But there's one crucial element to navigation that humans can do that computers currently can't - make arbitrary, seemingly illogical decisions in the heat of the moment. Take GP's hypothetical about the cliffside road - most humans, thanks to self-preservation instinct, will choose to rear-end the other car rather than drive off the ledge; what would a computer that is programmed to "avoid contact with other cars at all costs" do in that situation? Hyperbole aside, there's no way of knowing until we put one in that real world situation.

      Yes. Humans would never drive off a cliff, thanks to this self-preservation instinct. A self-driving car, on the other hand, could make such mistakes, because the engineers would never think of a scenario slashdotters come up with 5 seconds after seeing a story about self-driving cars. They would certainly never put their algorithms to the test. If it compiles, ship it. Right?

      Relevant quote from the last link:

      So confident is Volvo in this safety mechanism, it says that if the lead vehicle were to drive off a cliff, the next vehicle could stop before reaching the edge.

    168. Re:Must past this test by CanHasDIY · · Score: 1

      The concept that an auto-driver would be programmed to "avoid contact with other cars at all costs" and not have a concept of cliffs vs level verges is not a sensible one.

      I wouldn't think so, but IANA Auto-car programmer, so I couldn't begin to tell you what logic the thing functions off of. Speaking of which, is Google being very forthcoming with the source? That's something I wouldn't mind taking a look at myself...

      Well lets see. Apart from vision we have hearing, sight, touch, smell, taste, temperature, pain, balance, acceleration. Some of these are of little use in driving,

      You might be surprised - touch and spatial orientation are actually crucial to proper driving, almost as much as vision and hearing. Of course, being an 'old-skool gearhead,' I'm probably a bit more in tune with such things than a lot of other drivers on the road, so I suppose I'll have to concede that point on the law of averages.

      some of them are compromised in human drivers (e.g. soundproofing in modern cars, and the playing of music largely dicards sound as a useable sense by the average driver.

      There's a Buick/Caddy/whatever ad out there - it starts with a pan around a busy, noisy city street: construction workers with jackhammers, emergency vehicle sirens blaring, loud diesel trucks moving about. Cut to an inside shot of the new Buick Silencer (or whatever, not important), and the driver is smiling at the silent serenity of his cabin (at this point, the ad notes how quiet the interior of the vehicle is). Best part: at the end of the ad, they show the driver, content in his nearly silent reverie, pull off from the stop - directly into the path of an oncoming ambulance.

      2-fold moral for this story: 1) Sometimes 'quiet' is highly overrated, and 2) advertisers are freakin' morons.

      And yes, that was a ridiculously circuitous way of saying I agree with you.

      But of all the driver-useful ones, technology gives us better sensors then the human ones.

      OK, fair enough.

      Not being a designer of such systems I can't answer that. But it is a pretty obvious hazard, and can be assumed to be dealt with in a reasonable way. I do know that there is the principle of fail-safe and redundancy, so I'd expect a car with the loss of one of it's sensors to still be able to function safely, but to alert the human to the fault, and possibly to drive to a safe parking place and refuse to continue on automatic until the fault has been rectified.

      Hm, not sure about that one... the 'limp-home' mode makes sense, but I think due to liability reasons (we do live in a rather litigious society, after all) the machine would more likely switch over to manual control.

      I'd expect information fed via radio from other auto-drivers to be treated like the information a good human driver receives from other car's turn signals. Useful advisory information for maintaining a smooth drive, but not to be relied upon as totally reliable. So a safety margin is still left, regardless of what the other car actually does. If the information is wrong, I'd expect it to result in an emergency stop, rather than a collision.

      ... I wouldn't so readily discount the damage that could be wrought by a malicious individual, were I in your Bruno Magli's. I think time will tell on this one, it's not like we have a shortage of people who enjoy compromising computer systems for fun and/or profit.

      "When we least expect it, life sets us a challenge" - Paulo Coelho (Brazilian novelist)

      When I say "I'd expect" what I mean is that unless and until such systems behave like this or better, I wouldn't expect them to pass the safety tests and be licensed for consumer use. And I mean that I believe these things are possible with today's technology.

      Well, until we invent an auto-politician that functions purely on logic

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    169. Re:Must past this test by quacking+duck · · Score: 1

      Odds are cliffs do not move often and any automated car will have access to maps with topo data.

      Given the last directions I got from Google Maps concluded with 'now drive through the barrier at the side of the highway and fall forty feet into the parking lot of the hotel below you', that does not give me warm fuzzies.

      Wait, you mean maybe iOS6 replacing Google Maps might actually have been a *good* thing???

    170. Re:Must past this test by Anonymous Coward · · Score: 0

      I think your strawman argument that one'a them new-fangled awto-mobile self-drivin' horseless carriages would drive off a cliff instead of hitting the car in front of it.

      You know nothing about how they're programmed (maybe they sense they're in a lane next to a cliff?), and you should be sterilized for making such a stupid assumption and retarded comment.

      Asshat.

    171. Re:Must past this test by Cro+Magnon · · Score: 1

      Once, I was driving down the road at 35 MPH, and a car popped in front of me from a cross street! My street was two lanes, one going each way, and there was clearly no oncoming traffic, and I was too close to stop in time. If I hadn't swerved, I would have T-boned the other car.

      I suppose it's possible that a self-driving car could have stopped in time, but not certain. I wouldn't trust a car that was programmed to NEVER swerve.

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    172. Re:Must past this test by Anonymous Coward · · Score: 0

      Your scenarios a not even a little realistic.

      Every self-driving car for, oh, the first century at least, will have a mandatory human override, because sometimes humans are better than cars at driving.

      Your idiot who lost the privilege to drive won't be able to use a self-driving car, because he is not permitted to use the mandatory override.

    173. Re:Must past this test by Anonymous Coward · · Score: 0

      These questions have already been answered.

      You come to a complete stop on an icy road and put on the parking break. The incline of the road combined with the ice cause your vehicle to slide sideways and into another parked vehicle. Who is liable? In at-fault jurisdictions, you are liable. In no-fault jurisdictions, you are both liable.

    174. Re:Must past this test by cyberchondriac · · Score: 1

      :-)
      That could be beneficial, cars with a hive mindet, I suppose. If it's implemented right, it'd be like a large flock of birds in flight, suddenly changing direction or speed all at once, in unison, with never a collision. (It still amazes me sometimes that they can do that and not knock some of themselves out of the sky.) I guess schools of fish do that too.

      --

      Look back up at my post, now look back down, you're on the Internet. Now look back up. I'm a signature.
    175. Re:Must past this test by HiThere · · Score: 1

      That's actually being proposed as a system called Car-Talk, to allow cars to see around corners. Sounds reasonable. *Sounds* like it could be secure. Will it be? Who knows. Some idiot may decide to use a Turing complete protocol. But for now it sounds good.

      --

      I think we've pushed this "anyone can grow up to be president" thing too far.
    176. Re:Must past this test by Cro+Magnon · · Score: 2

      Does it at least suggest keeping your windows closed during your trip?

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    177. Re:Must past this test by Anonymous Coward · · Score: 0

      I think there were several issues including human factors in the design of the interfaces; but I really think that these guys got used to being along for the ride and it was not conceivable to them that the plane had decided to stop flying itself.

      A clear failure to remember Airbus' Golden Rules:
      Especially this one: http://www.youtube.com/watch?v=22Os-hzPZ7E&t=95s
      Also explained here: http://www.youtube.com/watch?v=pZnKevgNwu0&t=161s

      And especially: http://www.youtube.com/watch?v=pZnKevgNwu0&t=193s

      Entire videos here:
      http://www.youtube.com/watch?v=22Os-hzPZ7E
      http://www.youtube.com/watch?v=pZnKevgNwu0

    178. Re:Must past this test by Kaptain+Kruton · · Score: 1

      People keep mentioning things like faster response time and the ability to monitor more things. Yes, a computer can respond much faster and can take in more data. But will it respond to the right data? How often do you think the driving system is going to say, 'Hey these roads are covered in ice, I am on a multi-lane road, and the guy behind me is driving like a moron. I am going to purposely drive slower so the other guy changes lanes and passes me.'? I have done this.... and it was a good idea because I saw the same car unable stop when he arrived at the ice-covered intersection (luckily there was very little traffic). Good driving comes from experience and predicting what the other drivers will do.... not just taking in real-time data and responding to it.

      Yes there are some things that machine are better at than we are.... but driving is one of the things in which you constantly find exceptions to normal conditions and you cannot count on other drivers following the rules. If the rules are useless, then the abilities of the computer become much less effective.

    179. Re:Must past this test by Nimey · · Score: 1

      I suspect what happened with the airliner is this:

      Newbie junior pilot does't notice that he's getting bogus readings from instruments after he's disengaged autopilot when it goes crazy. Tries to fly by the seat of his pants and doesn't realize that he's still descending because he's stalled without enough airspeed. Panic. Everyone dies before a more experienced pilot can take over and point the nose down to gain speed.

      --
      Hail Eris, full of mischief...

      E pluribus sanguinem
    180. Re:Must past this test by Kaptain+Kruton · · Score: 1

      An AI is only as good as its programmers. Programmers are human. They will make mistakes. They will get things wrong. They will fail to consider certain things that typically do not apply to the region in which they live. When these things happen, the only thing the precision sensor equipment will do is ensure the insurance companies and/or emergency response teams will be able to get precise information about what happened during the the crash.

      Also, I doubt the auto-drive system will be COMPLETELY separate from the other computing devices. Think about it... bugs in the software will be found and they can either create a way for the updates to automatically happen through the internet or some wireless means, or they can force the drivers to inconvenience themselves by taking it to a shop for the update (which would also cost the company money because they would pay for at least some of the technicians labor). Also, self driving cars need to have some idea of how to get from point A to point B. This information changes frequently due to construction/detours, unexpected events, and the creation of new roads... without this information being updated, the self-drive feature will become useless.

    181. Re:Must past this test by Kjella · · Score: 1

      Human drivers allow fatalities everyday. The question is not is it better than some hypothetical human driver, but is it better than the drivers we have right now.

      But when I get behind the wheel I don't care if it's better than some hypothetical average driver, I care if it's better than me. And since some 90% of drivers think they're above average drivers, they wouldn't replace themselves with anything less. Even if was a coin flip whether the car would kill me or I'd kill myself, I'd clearly pick being in control myself. That would be the consequence of my own poor driving, rather than being accidentally killed by a car. Particularly if the situation is against the statistic where a human driver would have survived while the driverless car fucked up. Those cases would clearly happen in a 50-50 situation, you'd win some and you'd lose some.

      To really take over you have to reach a level where most people accept that the driverless car is better than them, maybe the 90th or 95th or 99th percentile. That even if you're awake, sober and paying attention to the road you still don't have a 360 degree sensor grid, direct access to the car's systems like road grip, night vision, lightning-fast reaction skills and whatnot else a computer can do that you can't and the failures so few you can dismiss as real oddball cases you probably wouldn't have resolved well yourself either. On the other hand, any car approved for the road would hopefully get rid of a few drivers that probably know they shouldn't be driving but are holding on to their driver's license for their dear life.

      --
      Live today, because you never know what tomorrow brings
    182. Re:Must past this test by JTsyo · · Score: 1

      well then driving off the cliff solves almost all his problems.

    183. Re:Must past this test by JTsyo · · Score: 1

      I could see the cars connected to each other to pass information or a highway network that centrally control traffic.

    184. Re:Must past this test by Anonymous Coward · · Score: 0

      Ok, fine, then i don't want my taxes to pay for your stupid ass utopia.. or your health care either.

    185. Re:Must past this test by h4rr4r · · Score: 1

      I only ever drive manual. Still boring if the drive is boring.

    186. Re:Must past this test by MrBigInThePants · · Score: 1

      What complete rubbish. The road rules would be the same.

      All that is required is that one person in the car is designated the "operator" (i.e. chooses where the car goes) and they are recognised as holding full responsibility for the car.

    187. Re:Must past this test by Anonymous Coward · · Score: 0

      An AI is only as good as its programmers. Programmers are human. They will make mistakes. They will get things wrong. They will fail to consider certain things that typically do not apply to the region in which they live. When these things happen, the only thing the precision sensor equipment will do is ensure the insurance companies and/or emergency response teams will be able to get precise information about what happened during the the crash.

      An AI is only as good as its inputs. The programmers are not the only input; there is also the example driving. I am not sure what combination of hard-coded driving rules and machine learning Google is using, but you can be sure that's there's some of both. This doesn't really break your point: the inputs are still chosen by humans and necessarily incomplete.

    188. Re:Must past this test by KingMotley · · Score: 2

      No, the question is: is it better for OTHER people to have them than let them drive?

    189. Re:Must past this test by LeadSongDog · · Score: 1

      I'm a little confused. Can you restate that with a car analogy?

      You're mixed up, like "a red VW Microbus, with shovels, and rakes, and inference by induction."

      --
      Oh, I'm sorry sir, I thought you were referring to me, Mr. Wensleydale.
    190. Re:Must past this test by KingMotley · · Score: 1

      More than likely, you'll pay higher insurance premiums based on the quality of the AI and it's history of accidents.

    191. Re:Must past this test by Anonymous Coward · · Score: 0

      Driving a car on a public road isn't a right.

      Actually, it's a right. Travel in general is. If it were a privilege, states could restrict who drives and who doesn't without any due process.

    192. Re:Must past this test by Anonymous Coward · · Score: 0

      I assume there will be cameras and courts will decide whether it was a true accident (very few of those, in fact some agencies now use the term collision instead to avoid giving a false indication of the scenarios) or a bug in programming of one car or another.

    193. Re:Must past this test by Anonymous Coward · · Score: 0

      ... I'm not sure I see a problem here.

      Are you saying that humans should continue to suffer and die so that auto-insurance companies won't need to adapt / disappear? Are you saying that paying auto-insurance is a good thing at all? Are you saying that there isn't a way for insurance companies to ding the software-writers for payment?

      Are you even sure that driverless cars will have accidents at all, say, 20 years from now? Or that cars will be so expensive / non-disposable that insurance will need to exist? You know that computer-driven cars are all but certainly going to lessen the severity of car wrecks that do occur, right? 360-degree vision, light-speed reflexes, advances in materials technology for lighter frames and engines (kinetic energy), light-speed cooperative communication with all other vehicles in the area...

    194. Re:Must past this test by Anonymous Coward · · Score: 0

      Don't forget the human eyes' blind spots which is sometimes disastrous.

    195. Re:Must past this test by Anonymous Coward · · Score: 0

      I don't really think we need a list of his hobbies for that.

    196. Re:Must past this test by Anonymous Coward · · Score: 0

      When I see an animal on the road, I speed up and hope my aim is good.

    197. Re:Must past this test by slick7 · · Score: 1

      No that question is; Is the car a better driver than me when I am sleep deprived, upset at my wife and in a hurry to get home?

      The computer will always drive the same, humans are not the reliable.

      The real question is... how soon will the proximity detonator be developed?

      --
      The mind conceives, the body achieves, the spirit manifests.
    198. Re:Must past this test by Anonymous Coward · · Score: 0

      But since we have not mastered true AI yet

      We haven't mastered Natural Intelligence yet, either. Nor will we *ever.*

      But people can still get licenses to drive.

    199. Re:Must past this test by xaxa · · Score: 2

      1) says some anonymous person that doesn't know me from Adam. Anyone capable of cogent thought will take your statement with a heaping serving of NaCl

      Look at the timeline at the bottom: http://www.drive.com.au/Editorial/ArticleDetail.aspx?ArticleID=56781

      The airbags have deployed and the crash is over before the human notices.

    200. Re:Must past this test by xaxa · · Score: 1

      Hey these roads are covered in ice, I am on a multi-lane road, and the guy behind me is driving like a moron. I am going to purposely drive slower so the other guy changes lanes and passes me.'?

      That sounds like an easy rule to implement.

      In general, a following driver going, say, 15% faster than what the car computer considered a safe speed on that road (just a second ago) could be a good catch-all for bad drivers.

    201. Re:Must past this test by CanHasDIY · · Score: 1

      1) says some anonymous person that doesn't know me from Adam. Anyone capable of cogent thought will take your statement with a heaping serving of NaCl

      Look at the timeline at the bottom: http://www.drive.com.au/Editorial/ArticleDetail.aspx?ArticleID=56781

      The airbags have deployed and the crash is over before the human notices.

      And of course, anecdote = fact!

      How silly of me to not believe everything I read on the internet!

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    202. Re:Must past this test by Luke_2010 · · Score: 1

      For each one taking a driving license there are high stakes someone will hit a car sometime in his life. How do insurance companies handle that? They do. For being a better option self-driving cars just have to be better than us, they don't have to be perfect. Besides, technology improves rapidly, so in a short while they'll become almost perfect. We won't. No matter how long you wait for it to happen.

    203. Re:Must past this test by lister+king+of+smeg · · Score: 1

      no, my grammatical error, what i meant to say was if you do those while driving you are such. i saw that after i posted but didn't have time to post to correct it until now

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    204. Re:Must past this test by mantissa128 · · Score: 1

      Nirvana fallacy. The cars don't have to be perfect, just better than human drivers in most circumstances.

    205. Re:Must past this test by wonkey_monkey · · Score: 1

      Basically what I am waiting for is the inevitable 100 car pile up with massive fatalities that WILL occur at some point in time where investigation will identify that a self-driven car, or cars, was the cause of it.

      A 100 car pile up sounds to me like evidence of at least 90 cars crying out for AI drivers.

      --
      systemd is Roko's Basilisk.
    206. Re:Must past this test by cheater512 · · Score: 1

      Yes

    207. Re:Must past this test by Zalbik · · Score: 1

      What's wrong with the suggestion? Step 41. clearly says: "Jet Ski across the Pacific Ocean"

    208. Re:Must past this test by geoskd · · Score: 1

      I fly a small plane for fun. I'm certainly more dangerous up there than a military or airline pilot, and it's far less efficient.

      At some point, you have to let people do what they're going to do even if there's some measurable risk to the public. It's a fine balancing act sometimes, but it's also part of a free society.

      If driving required the same degree of training, and was as controlled as flying then I don't think it would be an issue. The problem is that they give *every* idiot 17 year old, who asks for one, a license. Now, driving on a closed track on the other hand, is something entirely different. In short, you and your plane pose orders of magnitude less risk to the public than one driver on a cell phone...

      On a side note, try flying your little single engine plane into commercial airspace, and not obeying ATC. Lets see how long it takes you to get shot down... If ground traffic had the equivalent of an ATC, there would be far less accidents. Also of note would be that if air travel involved the same kinds of volumetric density as ground travel, it would be virtually impossible for a human to pilot an aircraft safely.

      -=Geoskd

      --
      I wish I had a good sig, but all the good ones are copyrighted
    209. Re:Must past this test by MrL0G1C · · Score: 1

      An auto-driver will be better at maintaining a safe speed.

      I'd love to see this in action on the motorway when traffic is moving at a slowish pace, someone would cut in front of the computer controlled car, at which point it'd have to back-off to maintain correct distance... rinse & repeat.

      --
      Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
    210. Re:Must past this test by Anonymous Coward · · Score: 0

      Compared to your other recent comments, this thread seems a bit dumb, so I'm going to assume you're also in this time zone and wish you good-night!

    211. Re:Must past this test by Jedi+Alec · · Score: 1

      wouldn't you rather be reading, talking on the phone, texting, preparing for your work presentation/speech, doing your homework, etc?

      I thoroughly enjoy driving, so no, not really, but having the option available would be wonderful, especially if we reach the point where it's legal to have the car drive you home while under the influence.

      And no, there's no doubt in my mind that the AI will do a better job of driving than I will ever be able to. There's nothing like have actually having eyes in the back of one's head(or in this case, car) for making the job easier.

      --

      People replying to my sig annoy me. That's why I change it all the time.
    212. Re:Must past this test by tibit · · Score: 1

      Once you detect the need to take immediate action, you need to move your leg to hit the brakes.

      That's why automatic transmissions are better: you can keep your left foot on the brake pedal all the time. I've had silly arguments with people who thought that was somehow bad (because their daddy taught them differently, so they'll stick to it, ho hum). Driving with both feet in an slushbox car cuts your reaction time when braking, and allows better speed control when changing lanes, merging, etc. Like, duh.

      --
      A successful API design takes a mixture of software design and pedagogy.
    213. Re:Must past this test by BenFenner · · Score: 1

      A much more likely scenario is that the self driving cars prove statistically to be safer than human driven cars. At that point expect legislation to ban humans from driving.

      Can you explain why you think this is the more likely scenario taking into consideration that modern cars are statistically safer than previous generations (go back far enough and the differences are enormous) yet no US law I'm aware of has ever limited my ability to purchase and drive an older car and drive it legally on the road no matter how unsafe it may be compared to contemporary vehicles and regulations?

    214. Re:Must past this test by Jedi+Alec · · Score: 1

      As another situation, I am very interested in how these cars handle icy driving conditions.

      Modern cars already use ABS, TC and ESP to monitor the amount of grip the tires have and adjust braking and throttle, effectively overriding the driver, based on that.

      If I floor my 14 year old car in slippery conditions and the tires start to spin, it'll decide all by itself that the trottle needs to be reduced. By the time my vastly slower human senses have figured out what is going on, the whole thing is over already.

      --

      People replying to my sig annoy me. That's why I change it all the time.
    215. Re:Must past this test by tibit · · Score: 2

      What are those fantastic "other safety features that would considerably reduce deaths" that you claim? Care to elaborate? U.S. has been a leader in safety requirements for cars for quite a while I'd think.

      --
      A successful API design takes a mixture of software design and pedagogy.
    216. Re:Must past this test by SternisheFan · · Score: 3, Funny

      well then driving off the cliff solves almost all his problems.

      "I'm sorry Dave, but I had to disable your airbag. It's for the best, really."

    217. Re:Must past this test by Jedi+Alec · · Score: 1

      New roads that aren't in the GPS system, for example, or private driving areas (e.g., parking lots) that aren't well-mapped.

      Once these AI's take off, it'll only be a small amount of time before the signs next to roads will be modified or added to so automatic cars can "read" them and follow their instructions. Relying on GPS as the sole means of plotting and following a route would be ridiculous. Think roadworks, reduced speed limits due to all sorts of conditions, etc.

      --

      People replying to my sig annoy me. That's why I change it all the time.
    218. Re:Must past this test by tibit · · Score: 1

      I'd think that in areas where the radar coverage for ATC only uses transponder-reading radar, they may not even notice you ;) Not every civilian ATC radar is NORAD-worthy, you know.

      --
      A successful API design takes a mixture of software design and pedagogy.
    219. Re:Must past this test by Jedi+Alec · · Score: 1

      Highways? Highways aren't fun. Empty B roads are fun. Roundabouts can be fun.

      But hey, go ahead, decide what other people do or do not enjoy. I'll just go and have fun out there, within the rules of the road, while you look on disapprovingly. Deal?

      --

      People replying to my sig annoy me. That's why I change it all the time.
    220. Re:Must past this test by anubi · · Score: 1

      This oughta be a lot of fun for the tailgater!

      --
      "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]

    221. Re:Must past this test by tibit · · Score: 1

      You must be living in lala land if you seriously think that said account is not how those things are engineered to behave, and how they typically do behave, in fact. Hand in the geek card. Do not pass go. Sigh. That's why driving an old car is insane if you care about your survival and injuries. Even an otherwise decent Volvo from early 1990s doesn't do half of it, yet it's standard on today's cars.

      --
      A successful API design takes a mixture of software design and pedagogy.
    222. Re:Must past this test by tibit · · Score: 1

      Systems engineered for safety, like a self-driving car would, come with a book of requirements. A thick book, with thousands of numbered paragraphs. Sure the humans who set it all up could have screwed up, but tell you what, if you think various hazardous situations were not given due thought and analysis, you're deluded. They have likely thought of things neither of us will ever experience on a road, I bet on that.

      --
      A successful API design takes a mixture of software design and pedagogy.
    223. Re:Must past this test by Anonymous Coward · · Score: 0

      The wife is at work!

    224. Re:Must past this test by tibit · · Score: 1

      I wouldn't think so, but IANA Auto-car programmer, so I couldn't begin to tell you what logic the thing functions off of. Speaking of which, is Google being very forthcoming with the source? That's something I wouldn't mind taking a look at myself...

      Shows that you don't know much about properly engineered safety relevant software. Source is almost useless without all the engineering process documentation behind it. Heck, lack of such engineering process pretty much means the source is junk. Reverse engineering existing source to reproduce all the documentation is usually more work than doing it from scratch but properly this time. You must have requirements, designs, source, tests, validation results and full traceability between all of those. A relatively simple product might have a good 500 to a 1000 pages worth of documentation behind it (full pages). Something like a self driving car would be documented in 10s of thousands of pages, of course with help of automated tools to manage said documentation -- no one checks traceability by hand in a project of that size.

      Everyone on the team, including the management structure and responsibilities must be laid down and tracked throughout project/product lifetime. When you have process like that, you can have programmable electronic systems whose failure would be catastrophic, yet you count on them every hour of every day and they fail no more often than designed for, usually less than that. Good luck to you if your ABS computer decided to modulate your brakes into oblivion, or your stability augmentation computer decided to lock all four of your wheels for good measure. Never mind software that will you know drive the whole fucking car.

      --
      A successful API design takes a mixture of software design and pedagogy.
    225. Re:Must past this test by mug+funky · · Score: 1

      a voice said "use the force, let go". so i turned off the computer.

    226. Re:Must past this test by BasilBrush · · Score: 1

      So what you are saying is that you drive closer than is safe to the car in front in order to stop other cars going into your lane.

    227. Re:Must past this test by tibit · · Score: 1

      Huh? To be safe, any self-driving system will have a pre-analyzed path that it keeps updated as it goes. The speed is adjusted continuously to keep this path ahead of the braking distance. When it doesn't know what to do, it has already stopped. It can beep to alert you that you might need to get your car out of the way.

      --
      A successful API design takes a mixture of software design and pedagogy.
    228. Re:Must past this test by BasilBrush · · Score: 1

      Discussions like this are how we figure out what might go wrong, what to test for, which I for one am happy to be part of. Better to discuss the potential outcomes like adults, than engage in the proverbial dick-swingin' contests and hyperbolic nonsense that does no good for either camp.

      If we were in the business of designing such systems, sure. But any potential problem thought up by a non-domain expert on Slashdot will be old news to the people actually designing these systems.

    229. Re:Must past this test by tibit · · Score: 2

      Rimshot :) Such resources already exist in mature organizations like NASA. They have a lessons learned database that is pretty much required reading for any engineer who doesn't want to stagnate.

      --
      A successful API design takes a mixture of software design and pedagogy.
    230. Re:Must past this test by Anonymous Coward · · Score: 0

      Hi brand-new-account anti-Google first post.

    231. Re:Must past this test by tibit · · Score: 1

      Ha ha ha. You must have been asleep and not following state of the art in man-machine interaction, then. Keeping the human out of the loop, in various scenarios, has been shown to be potentially very dangerous and has been linked to mishaps and accidents in aerospace, naval and industrial operations. Automation that makes human's job overly easy, as is the case for example with modern glass cockpits in aircrafts, has been shown to be, counterintuitively, detrimental to safety of the overall man-machine system. The "rapid manual override" is precisely what seems to be a myth when you look at accident and mishap investigation reports. It just doesn't happen when human is not in the loop.

      --
      A successful API design takes a mixture of software design and pedagogy.
    232. Re:Must past this test by MrL0G1C · · Score: 1

      No, I cycle so I have to leave a safe distance to the car ahead, some drivers perceive a gap between me and the car ahead as a place to overtake into, pure stupidity as in slow moving London traffic it will probably be be mere seconds before I can overtake again.

      --
      Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
    233. Re:Must past this test by tibit · · Score: 1

      A good driver will notice a dangerous situation and take manual control well before a bad event happens.

      Reality would tend to disagree with you. Humans don't work that way. Your fantasized good driver doesn't exist. Sorry to burst your bubble, but accident and mishap reports plainly and simply say otherwise. Yeah, it's counterintuitive perhaps, but you have to get over it and accept the facts as they are. There are things we, as humans, no matter what existing training is applied, are seemingly unable to do. At least at this point in time. Perhaps in the future training will improve to tweak our brains to deal with such scenarios appropriately. So far, the consensus is that we the humans are broken as soon as we're out of the loop. This is also a contributing factor to why big corporations seem often like Titanics going full steam into the iceberg. The higher-ups are so far removed from the operational reality that they are demonstrably unable to take over when it's time. Apparently, the much longer timescales involved are not the matter at all.

      --
      A successful API design takes a mixture of software design and pedagogy.
    234. Re:Must past this test by Anonymous Coward · · Score: 0

      To steal a line from the right-wing nutjobs playbook "I'll keep my clutch, my wheel, and my car; you can keep the change."

    235. Re:Must past this test by Anonymous Coward · · Score: 0

      1) You reaction time is far worse than a computer.

      True. On the flip side you can better anticipate other drivers actions and increase safety margins around cars that drive oddly. You are also better equipped to handle unexpected situations.

      2) Your estimation of distances is far worse than machines absolute measurements.

      Unless the sensors fail. Getting obscured by snow or mud is a major concern. Then again human vision tends to worsen with age.

      3) You are limited to two forward facing eyes, augmented by 3 small mirrors. And you share some of the vision time with looking at the dash. An auto-car can look in all directions at once, and monitor all dashboard information and more at the same time.

      Untrue. While the computer can switch between contexts far faster than you can, it still isn't checking all of them at the same time.

      4) An auto-driver will be better at maintaining a safe speed. Able to stop in the distance it knows to be clear far more often than a human driver.

      Is this such an improvement? Humans deal with the inaccuracy by stopping earlier than they need to, an effective strategy. The higher accuracy of digital sensors is a great asset, but that doesn't mean human vision is worthless.

      5) I'd expect an auto-driver system to be seperate from any other computing devices in the car, and connected to the internet or any other vector for hacking. I'd expect them to be as immune to hacking as an auto-pilot system in a plane.

      Ha ha ha. One can certainly hope, but deep down in your soul you know the systems will eventually be merged together.

      I for one welcome our new self-driving overlords, but that doesn't mean I believe they'll immediately be perfect replacements for humans. I don't doubt they'll be safer in 95% of circumstances, but I think I'll likely wait until my current transport wears out before upgrading.

    236. Re:Must past this test by bkk_diesel · · Score: 1

      As someone who has a driver, I can tell you that I probably get the urge to take the wheel about once a month. Usually in the morning on a nice day when I want the music loud. FWIW, I would never let my driver sit in the back though - he gets the passenger seat.

    237. Re:Must past this test by mellon · · Score: 1

      Hopeful. People who are younger think that they are immortal, and will always have good reflexes; as you get older you realize that at some point you're going to be a danger to society if you keep driving. Self-driving cars will help a _lot_ of people, including your parents and, later on, you. And if they get bad drivers off the road, that will help even more people, because motor vehicle accidents are a _really_ major cause of preventable death in the U.S.

    238. Re:Must past this test by mellon · · Score: 1

      Equal protection under the law is a principle that's been abused pretty badly, but it is codified in the constitution. If you let anyone drive, then in principle you have to let everyone drive, unless there is some generally applicable rule that, if broken, would have the result that anyone who broke it lost their driving privilege. As long as that's the case, equal protection under the law is being observed. In order to observe it, there have to be legal proceedings. So that's why your driving privilege can only be revoked with due process.

    239. Re:Must past this test by mellon · · Score: 1

      No problem; the car pulls over and requests assistance. Probably the guys at OnStar take over and drive your car home for you by remote control.

    240. Re:Must past this test by Anonymous Coward · · Score: 0

      Now let's not drag Apple Maps into this...

    241. Re:Must past this test by drerwk · · Score: 1
      When I was a private pilot I subscribed to NTSB Reporter. I like this review of the journal:

      By Jim Carson Subscription Term Name:1 year By nature, this is a very depressing publication: Peter Garrison reviews aviation accidents that usually resulted in someone dying. One subscribes NTSB reporter in hopes of avoiding similar mistakes. A shorter version of Peter Garrison's NTSB columns appear in Flying Magazine. NTSB Reporter's trademark is a *much* more thorough analysis of a specific accident. It's immensely thoughtful, but I would not leave copies lying around on the coffee table for non-pilots or spouses. to see them.

      I suppose that the average recreational pilot takes more care than average car driver, yet the NTSB reporter can be summed up as follows - 1) forgot to gas up the plane 2) did not check the weather report 3) was in a hurry 4) the other 1% of accidents.
      I do not think any driver is good enough to always pay attention - unless they simply turn the robot off.

    242. Re:Must past this test by Anonymous Coward · · Score: 0

      It would help if you had the first clue what you're talking about. These "delusional" governments aren't simply allowing driverless cars on the roads to move around unrestricted. There are specific requirements protected by law governing safeguards such as there must be a person behind the wheel at all times, able to take over control of the car immediately. Whenever any kind of technology like this occurs for the first time, there has to be a field test, or an actual point in time when we take a calculated risk in order to objectively assess the technology in a real world environment. That time has apparently arrived. Or should be just forget progress because it might bring a small risk, and shove it on a shelf for the next 1,000 years because we're too scared to ever allow it to be tested.

      If it were up to you, we'd have no cars, because those things were dangerous an a menace to horse riders back in the day, no planes, because actually allowing one to fly near an urban area would be just inviting trouble, and no space travel, because, well... just thing of the dangers involved in that. We'd all be walking around (slowly to avoid potential tripping hazards) and would probably never leave our caves.

    243. Re:Must past this test by Anonymous Coward · · Score: 0

      We all are paying for that road, asshole.

    244. Re:Must past this test by Anonymous Coward · · Score: 0

      The human needs to be keeping a map of all nearby cars, just as the AI would. This map really does not need to get updated at superhuman speeds. Just pay attention!

    245. Re:Must past this test by Lucractius · · Score: 1

      This sort of 'flocking' and 'schooling' behavior like birds and fish do, is pretty well understood and I would expect some sort of standard would develop pretty quickly to provide the same sort of 'mob behavior cues' that animals have that enable them to decide when they should follow versus when they should make their own decision.

      Like for instance all the cars pulling over towards the edge of the road together to let through the fire engine or ambulance on the way to respond to an emergency.

      --
      XML - A clever joke would be here if /. didn't mangle tag brackets.
    246. Re:Must past this test by T-Bucket · · Score: 1

      Uh, I hate to tell you, but I did pay for that road. Taxes, and all that.

    247. Re:Must past this test by TranquilVoid · · Score: 1

      The overall experience is still part of driving. Being in control (i.e. the driver) is comforting and that's what makes us ignore statistics, like shark attacks vs crossing the road*.

      Of great interest would be culpability in the event of two self-driving cars crashing. If you treat them as decision makers, like a human driver, then who gets the fine/gaol-time when it decides to turn in front of an oncoming car? Technically everyone is a passenger. Is it the owner's fault for choosing that model of software? The company's? Will the software have to be open source so it can be investigated?

      * The basis for this one is never clearly shown.

    248. Re:Must past this test by TranquilVoid · · Score: 2

      Much like choosing the floor at which the lift will stop makes you responsible for any of its software failures?

    249. Re:Must past this test by a_hanso · · Score: 1

      'Why did you turn off the computer...?

      Because he decided to use the Force?

    250. Re:Must past this test by TranquilVoid · · Score: 1

      Cross-communication would open up opportunites to reduce congestion and efficiently route traffic. Would be slightly annoying when you see you have been chosen to take the long way though.

    251. Re:Must past this test by Neil+Boekend · · Score: 1

      Decent robotcar software should not have simple rules like "Avoid crashing into cars at all cost". It should more be like "Choose the course of action with the least chance of death, with equal chances of death choose the course with the least injury, with equal chances of injury choose the course with the least chance of damage to the car".
      I am not a programmer (certainly not in the league of Google), so I couldn't design a program that could do this. However, in complexity I think the guessing of a human drivers next move is more complex than whether the area next to the road is a cliff or a drivable field, and thus the estimation of the risk the cliff poses should be easier than the estimation of the risk of crashing into the car. The software has to calculate the risks of all the courses of action and choose the least risk.
      However it could be that our computers aren't fast enough to calculate all the risks. I could imagine that it's possible to have a map that has a layer with damage risks: ie a steep cliff has a 100% risk of damage to the car, 99% risk of injury, 80% chance of death. This means the car would not take a dive off the cliff if the chance of death is less than 80% in the other case, which is unlikely to happen.

      Now there will be errors in the risk estimations. Off course there will be. But the main question is: will there be fewer errors in the software than in humans? I think there will be. But I'd require proof. 725,000 miles without accidents sounds like a reliable proof. I assume these are miles in the USA. Here in the Netherlands I'd require less, more like 100,000 km (62000 miles) since almost nobody gets further than 10.000 km without a fenderbender in the "Randstad". The roads are crowded here.

      --
      Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
    252. Re:Must past this test by Neil+Boekend · · Score: 1

      If the driverless car is available then it's possible to change road law to make it easier to ban people from driving and increase the age at which a drivers license can be obtained. All things said and done, for many people there simply is no alternative.
      These are good things.
      A decent marketing department would advertise this at least as a way to get home when drunk without the risk of driving over some kids. "Think of the beers, err... Kids." They would probably find a better slogan.
      And do you think Google wouldn't shove some campaign contributions to the people in charge if they promise to change laws to make it easier to ban people from driving? Its not evil to prevent deaths isn't it?

      --
      Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
    253. Re:Must past this test by Anonymous Coward · · Score: 0

      I'm pretty sure the insurance companies won't be a problem. Apart from the obvious (assuming fewer accidents overall, otherwise we wouldn't switch to self driving cars), self-driving cars don't mean the rules need changing. The current way they work is that they try to determine who is at fault. If no-one can be identified, the bill is split or everyone pays their own damages.

    254. Re:Must past this test by blippo · · Score: 1

      Why would a self driving car ever drive off a cliff?
      Clearly it would rank available options and pick the lowest cost one.

      Yes, in a science fiction fantasy world where True AI is working according to Asimov's laws.

      In the actual world, real persons are going to outperform computers in any situation that's "tricky" to analyse,
      and that will continue to be the case until the cars would be able to pass the Turing test.

      The cars can be given better sensors (ir, radar, etc) and faster computers to give the algorithms an advantage,
      but when it fails, it will fail in a way that a human would think is incomprehensibly stupid.

      When the first autonomous car crashes into a kindergarten because it did something stupid, it will be hard to point
      to the statistics and say that in the big picture, we're saving lives.

    255. Re:Must past this test by burning-toast · · Score: 1

      I shit you not, those directions indicate you should jet-ski 782km across the pacific ocean:

              40. Turn left onto 350

      3.4 km
              41. Jet ski across the Pacific Ocean

      782 km
              42. Continue straight onto

      - Toast

    256. Re:Must past this test by burning-toast · · Score: 1

      Damn slashdot... ate the chinese/japanese characters in the road names...

    257. Re:Must past this test by blippo · · Score: 1

      Hi.

      Before you argue more about this, write a program that does something really simple, like packing a few arbitrary objects in a box just big enough to fit them, and then come back to this discussion. ( I mean the task must be really simple, because it's something even a small child can do. )

      See you in a couple of years :-)

    258. Re:Must past this test by blippo · · Score: 1

      Are you sure you are not confusing reality with something you saw on tv?

    259. Re:Must past this test by BasilBrush · · Score: 1

      Are you sure you are not confusing reality with something you saw on tv?

      Tell me which of those things you think isn't possible, and we'll talk about it.

    260. Re:Must past this test by Ginger+Unicorn · · Score: 1

      Any more ad hominems or non sequiturs you care to throw out there?

      That's exactly what I thought when I read your point 1, an ad hominem, and point 2, a non sequitur.

      you're spectacular :)

      --
      (1.21 gigawatts) / (88 miles per hour) = 30 757 874 newtons
    261. Re:Must past this test by greenfruitsalad · · Score: 1

      I'm just waiting for the time when prioritisation gets implemented into this. e.g.: a politician in car A and a housewife in car B both heading in the same direction, at the same speed and next to each other. Something suddenly happens in front of them and there is only room for one car to avoid the obstacle safely. Big Brother decides who gets to live.

    262. Re:Must past this test by cyberchondriac · · Score: 1

      Wrong, asshole, I was doing 75

      --

      Look back up at my post, now look back down, you're on the Internet. Now look back up. I'm a signature.
    263. Re:Must past this test by CanHasDIY · · Score: 1

      I second that a computer has a faster reaction time than you do. Sorry, man, you seem outnumbered on this one.

      Well. first someone has to actually make that comparison, which OP did not; they merely stated "your reaction time is absolute crap," not "your reaction time is absolute crap compared to a computer." I wholeheartedly agree with the latter.

      P.S. y'all can keep the "dur, it's implied" to yourselves - I don't assume implication in the words of others, as that would require a belief in clairvoyance, which I do not believe in. Say what you mean.

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    264. Re:Must past this test by CanHasDIY · · Score: 1

      You must be living in lala land if you seriously think that said account is not how those things are engineered to behave, and how they typically do behave, in fact.

      I don't disagree with that, what I disagree with is the concept that A) a person "isn't aware" they are about to get smashed in the driver's door (I contend that if you're paying so little attention to your driving you don't see a truck barreling down on you, you probably don't need to be behind the wheel anyway), and B) that this is a universal truth.

      The technical aspects of the post, i.e. times of sensor activation/system deployment, are entirely accurate.

      Probably should have pointed out the specific aspect I found dubious from the get-go. Sorry 'bout that.

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    265. Re:Must past this test by CanHasDIY · · Score: 1

      Any more ad hominems or non sequiturs you care to throw out there?

      That's exactly what I thought when I read your point 1, an ad hominem, and point 2, a non sequitur.

      I don't think you really know what that means.

      you're spectacular :)

      Quite, thanks for noticing!

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    266. Re:Must past this test by Shrike82 · · Score: 1

      This is what the vast majority of anti-automated car people forget: an automated car will actually drive in a way that results in a net gain to the system. Everyone benefits when people leave a decent gap between their car and the car in front (go read if you don't believe me). A human driver often won't because they want to "send a message" to the other driver that "they're too slow" or they don't want someone entering "their" territory and will actively move to block them - I see this every day. Automated cars have the advantage of not giving a shit about emotional considerations, and they won't instantly turn into arseholes the moment they get into the "safety" of a car like a lot of the people who seem to share my commute each day. Apologies for the large amount of quotation marks; they're meant to indicate me using an annoying air quotes voice.

      --
      You can advertise in this sig from as little as £99.99 a month!
    267. Re:Must past this test by dave420 · · Score: 1

      Yeah, you're right. The cars just have webcams to see the road paintings, and assume there is nothing but flat gravel traps either side of all roads for at least 200'. The car in your first scenario would do what any normal human driver would be instructed to - slow the fuck down and try not to crash into anything. It would use its better senses to first detect the asshole car start to affect its safety, then do what it can to get out of the way, all with 360-degree vision (using a laser range-finder and other equipment), so it probably knows better than you how deep the ravine is and why it's such a bad idea to go driving in it. The people who make these things aren't idiots. They're actually more clever than you seem to give them credit for.

    268. Re:Must past this test by cellocgw · · Score: 1

      That's why automatic transmissions are better: you can keep your left foot on the brake pedal all the time.
      False conclusion from this fact. It's quite possible (and I've lost track of the biological studies) that you can move your active leg faster than the one you've been completely ignoring. The premise is that your brain can modify the control of the right leg faster than it can "wake up" control of the left leg. So, until some careful reaction time studies are done, I remain skeptical that "left leg on the brake pedal" is a good idea.

      --
      https://app.box.com/WitthoftResume Code: https://github.com/cellocgw
    269. Re:Must past this test by Matimus · · Score: 1

      It is legally defined as a privilege. States do restrict who drives and who doesn't. You have to obtain a special license from the government in order to drive a vehicle on public roads. If you are found to abuse your privileged then that license can be revoked.

      Effectively you are restricted from driving from birth without due process. You must go through a process in order to gain the privilege.

      You have a right to freely travel, that doesn't mean you have a right to a specific form of transportation.

      --
      GENERATION 25: The first time you see this, copy it into your sig on any forum and add 1 to the generation. Social exper
    270. Re:Must past this test by tibit · · Score: 1

      How is that left leg "sleeping"? In city traffic you use brake all the time, and even in congested highway traffic you use it at least once a minute. I've done some tests on myself and while I didn't look at situations where the brake is dormant for minutes or more, in repeated use it shaves off in the ballpark of 100ms from my reaction time, and that's conservative. If you're on an empty highway with cruise control enabled, then there's no difference in how "engaged" either leg would be. They both do nothing for long periods of time, so there's no difference between the legs. Ergo, you still save time on not having to move the right foot to the left pedal, because otherwise they are both just as slow in reacting.

      Two foot driving also permanently solves the problem of slamming the wrong pedal, because you never ever use right foot to brake, so your learned and trivial response to the perceived need to brake is to slam the left leg. You also mostly avert the problem of pressing two pedals at once if you mistarget the foot's landing, or having your foot caught between the pedals as you move it sideways. Not that pressing two pedals at once is tragic like some dumbasses out there would want you to believe.

      --
      A successful API design takes a mixture of software design and pedagogy.
    271. Re:Must past this test by cellocgw · · Score: 1

      You're an educated /. reader -- you must know that "in the ballpark," especially when applied to one's self, is nonsense. I repeat: wait for a prospective, null hypothesis test.

      --
      https://app.box.com/WitthoftResume Code: https://github.com/cellocgw
    272. Re:Must past this test by CosaNostra+Pizza+Inc · · Score: 1

      They sure try. You've seen the TV commercials about dead relatives who text'd while driving, right? Also I forgot to add sleeping, drinking (alcohol), sex, and anything else you can think of doing in a car other than driving.

    273. Re:Must past this test by Anonymous Coward · · Score: 0

      You're right, you were entirely justified in claiming that your reaction time is NOT crap (compared to a two-toed tree sloth). When the entire topic of discussion was that computer drivers would be better than human drivers, nobody would expect to compare human drivers to computers when making a claim about reaction times. That would be absurd.

      I leave you this in parting. Come back when your reaction times are not crap.

    274. Re:Must past this test by Mike+Buddha · · Score: 1

      Not for use at all times. When you're crusing on the freeway and driving is uneventful, turn it on. When you're on a windy dirt road with a sheer cliff on one side and a rock wall on the other, turn it off. This ain't rocket science.

      --
      by Mike Buddha -- Someday the mountain might get him, but the law never will.
    275. Re:Must past this test by Mike+Buddha · · Score: 1

      If I have a hammer I have to use it for every single task I have, otherwise what's the point of owning a hammer?

      I'm a little confused. Can you restate that with a car analogy?

      If I have a car that hammers nails into wood I have to use for every single task I have, otherwise what's the point of owning a car that hammers nails into wood?

      --
      by Mike Buddha -- Someday the mountain might get him, but the law never will.
    276. Re:Must past this test by pnutjam · · Score: 1

      100 mph club, here I come!

    277. Re:Must past this test by pnutjam · · Score: 1

      I see you've been watching me...

    278. Re:Must past this test by Martin+Blank · · Score: 1

      Shot down? Hardly. They just track you, maybe intercept you with a plane or helicopter, and follow you down to where police and the FAA are waiting. In any case, there's no such thing as "commercial airspace." What you're probably thinking of is the Class B airspace that's around major airports like LAX, San Diego, and Chicago. There's nothing prohibiting me from flying there so long as I have permission to do so. Class C airspace such as around John Wayne Airport in Orange County merely requires that they acknowledge me by tail number. So long as I hear at least, "Skyhawk 13974, stand by," I can enter the airspace. Likewise for the Class D airspace around smaller airports. Class E airspace requires even less communication--most of Southern California is encompassed in it and it's there primarily so instrument flights know where they'll have confirmed ATC coverage. I fly through it with zero contact after leaving the airport all the time.

      Besides, flying is hardly a "closed track." I practice over populated portions of the city of La Habra at altitudes varying from 1500 feet to 3500 feet. Stalls, steep turns, practice power loss, and all without any communications with ATC. Oh, I call out occasionally to let people know where I am, but that's strongly recommended, not mandatory. I do it primarily in case I'm in someone's blind spot. If I lose complete control, I'm probably coming down on someone's house. Individually, I may not pose as much threat, but what I do is more dangerous on a per travel mile to me and to my passengers, and maybe to people on the ground.

      --
      You can never go home again... but I guess you can shop there.
    279. Re:Must past this test by Martin+Blank · · Score: 1

      I should hope they're already there. After all, if the computer's maps say that the speed limit is 45 and the town has recently lowered it to 35, it should be able to read the sign and take that number over what it knows. New streets could still present a challenge, though.

      Besides, I didn't say that GPS was the only means. They have cameras and other sensors for a reason. But navigating a parking lot is not the same as navigating a road and has different challenges.

      --
      You can never go home again... but I guess you can shop there.
    280. Re:Must past this test by fireylord · · Score: 1

      Someone has never, ever taken an AI class. Or even an algorithm class dealing with risk. Here's how the calculation actually works (and by the way, that approach is about 20-30 years old).
      Every situation is assessed an impact value: driving into oncoming traffic, 0 (very bad); driving into the right ditch, 10; swerving into a legal lane, 50; etc. Every situation is given a set of possible actions, with each action having a probability of being completed successfully. The algorithm multiplies the outcome with the odds of achieving that outcome, and picks the highest value. You can set it up in different ways, but the idea is the same: multiply outcome severity with odds of achieving outcome, pick lowest combined risk/outcome. In your situation, driving off the cliff (which is assumed to be very bad, since the car can see a very steep drop-off with no bottom) is going to have a much worse outcome than hitting the car in front of it. Hitting the car in front of it is guaranteed, but so is driving off the cliff. As a result, the algorithm will make the automated car hit the car in front of it, rather than drive off the cliff.

      Not to mention that cars don't sleep, always behave optimally (according to the algorithms in place), and have no blind spots.

      Oddly, one thing that seems to not have been considered is that an AI car would have a much much better ability to handle the car well under heavy braking than a human. Surely the 'drive into the oncoming car' outcome would be somewhat offset by the fact that the computer would pretty much instantly apply the maximum braking force it thinks the ABS equipped car can manage whilst performing a manouvre, at the same time as performing it, thus massively reducing the energy of any impact?

    281. Re:Must past this test by Anonymous Coward · · Score: 0

      People have been accepting that since the first automatic carding and spinning and weaving and sewing machines first got good enough to make stuff, then got better than the best humans.

      More recently, the machines took over planting and harvesting cotton as well. And self-driving tractors are already in the works.

      The thing is, lots of people think of driving as a high-level task, when it takes 0 creativity. As evidenced by the fact that computers can be programmed to do it.

    282. Re:Must past this test by Idbar · · Score: 1

      What I was trying to imply, is that your list is a bunch of items of why a computer may perform better. I thought distraction and concentration are one key point in the current driving environment. A computer, probably shouldn't get distracted by mundane activities, random events or being tired/frustrated. That seems to be a permanent in the list, and not just "intermediate". Did I misunderstood your list?

    283. Re:Must past this test by BasilBrush · · Score: 1

      Sorry, no, I completely misread what you wrote. You made a good point.

    284. Re:Must past this test by MrBigInThePants · · Score: 1

      Nice try at straw man.

      You don't have ownership or duty of care to the lift in any way, the buttons allow you to only go where the owners want you to go and (most importantly) no laws say you are responsible (which was the crux of my argument).
      Also lifts have nothing to do with cars.

      But apart from that you are spot on...

    285. Re:Must past this test by TranquilVoid · · Score: 1

      Yes, in a lift you are a non-owner passenger, but I suspect legally ownership of a vehicle is a non-issue. The driver is responsible for ensuring the vehicle is roadworthy, not the owner (although I'm sure that varies by region).

      For a self-driving car you are an operator, not a driver, although legally it may be the same. But in a self-driving car the operator is very close to being a passive passenger. Even if they have to take responsibility as the operator, any flawed decision by the software is going to be akin to a tyre blowout or brake cable snap - if they were reasonably maintained the driver is not at fault. Possibly even less so in the case of software as it would be considered a design fault rather than a random material failure.

      This is why I still think the lift analogy stands. The law if full of contradictory principles, so legally there may be different responsibilities imputed to building owners and car owners, but if lift software makes the lift fall a floor, that's going to come back to the lift company, and similarly for car software. Were the drivers of the sticking Toyota accelerator held responsible?*

      *assuming it was clear the fault was the electronic throttle

    286. Re:Must past this test by Anonymous Coward · · Score: 0

      I think in this case we're determined that you are the problem and not the drive.

    287. Re:Must past this test by Ginger+Unicorn · · Score: 1

      Thanks for providing a citation to back me up. If you read that article it should help clear up your confusion. No link pointing to a definition of ad-hominem just to hammer my point home too?

      thanks for noticing!

      A brilliance this intense is impossible to ignore

      --
      (1.21 gigawatts) / (88 miles per hour) = 30 757 874 newtons
  2. They legalize driverless vehicles by Anonymous Coward · · Score: 0

    As long as said vehicles are not actually driverless.

  3. CA Freeways by sycodon · · Score: 3, Funny

    Having lived in CA and driven on the freeways, I can say that you don't need "self driving" cars for the freeways.

    All you need is a car that can self park and you are good to go...or...not go.

    --
    When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
    1. Re:CA Freeways by NoNonAlphaCharsHere · · Score: 5, Funny

      Anybody with any experience with California drivers knows that most cars in California are already "driverless".

    2. Re:CA Freeways by Anonymous Coward · · Score: 0

      Or hybrid muscle car - gasoline engine for the rear occasions where you can drive fast combined with bicycle pedals for the daily commute.

    3. Re:CA Freeways by Anonymous Coward · · Score: 0

      True, especially vans driven by moms.

    4. Re:CA Freeways by FatAlb3rt · · Score: 1

      I know you're joking, but a network of self-driving cars would probably prevent a lot of the gridlock as not only would follow distances be optimized, but they could follow much closer (since the array can communicate with one another so they brake uniformly).

  4. 300 million miles by Arthur+B. · · Score: 1

    3e8 car mile ~ 3000 cars * 1 year * 35 miles / hour * 8 hours / day * 365 days / year

    So if Google wants to reach that milestone, they need to start cranking out those self-driving cars.

    --
    \u262D = \u5350
    1. Re:300 million miles by h4rr4r · · Score: 1

      Why only 8 hours a day and why would you only ever drive it at 35 mph?

      A self driving car should be able to be on the road at least 20 hours a day. That leaves enough time for fueling and swapping of humans.

    2. Re:300 million miles by alphatel · · Score: 1

      You only need 1,000 cars to meet the milestone in 1 year, Driving at 30mph is the mean but 24 hours a day is plausible. And you don't need humans.

      --
      When the foot seeks the place of the head, the line is crossed. Know your place. Keep your place. Be a shoe.
    3. Re:300 million miles by h4rr4r · · Score: 1

      I believe on public roads you do need a human available to take over for legal reasons.

    4. Re:300 million miles by 0123456 · · Score: 5, Insightful

      I believe on public roads you do need a human available to take over for legal reasons.

      And that worked so well for AF447.

      Aviation autopilots should have proven by now that relying on a human to take over when the situation is so bad the autopilot can't handle it is a recipe for disaster. Besides, what's the point of a 'driverless car' if I have to be continually ready to take over at a millisecond's notice?

      Car: 'Warning, warning, kid just jumped out in the road, you are in control'.
      Driver: "WTF? I just hit a kid and smeared their insides all over my windshield'
      Car manufacturer: 'Not our fault, driver was in control, human error'.

    5. Re:300 million miles by h4rr4r · · Score: 2

      I agree Colgan 3047 backs that up even further.

    6. Re:300 million miles by localman57 · · Score: 1

      3e8 car mile ~ 3000 cars * 1 year * 35 miles / hour * 8 hours / day * 365 days / year

      So if Google wants to reach that milestone, they need to start cranking out those self-driving cars.

      So what are they basing this on? The number of miles a person would typically have to drive before they were "due" to have a fatal accident? And that is 300 Million?

      Maybe google should test driverless high-performance motorcycles instead. You'd only have to show you could go...what...maybe 1500 miles, assuming the computer isn't wearing a helmet? Having the computer ride next to one of his friend computers would probably make it even less. Also, you could have some cute girl computers pull up next to the motorcycle computer at a red light. Maybe a self-driving Kia Soul, or something like that. That would tend to cut the average down too.

    7. Re:300 million miles by camperdave · · Score: 1

      Do those have to be "real" miles, or can we just run them on a simulator?

      --
      When our name is on the back of your car, we're behind you all the way!
    8. Re:300 million miles by Arthur+B. · · Score: 1

      So what are they basing this on?

      According to http://www-fars.nhtsa.dot.gov/Main/index.aspx, the rate of fatal automobile accident in the US is about 1.11 fatal accident per 100 million vehicle mile.
      Assume the Google car's traveled distance between two fatal accident follows a Poisson law (that is there's a constant probability of having a fatal accident in a Google car for any mile traveled).

      Null hypothesis: the Google car has the same rate of accident as the U.S.
      The probability that the Google car given the null hypothesis has no fatal accident over 300,000,000 miles traveled is exp(-3.33) ~ 3.58%
      Thus the null hypothesis can be rejected with a p-value of 3.58%.

      To get a p-value 5%, the following would do:
      270M vehicle mile with 0 fatal accident
      430M vehicle mile with 1 fatal accident
      570M vehicle mile with 2 fatal accidents
      700M vehicle mile with 3 fatal accidents
      etc...

      --
      \u262D = \u5350
    9. Re:300 million miles by Anonymous Coward · · Score: 0

      So what do you propose the aircraft does when it receives invalid input from the wind speed sensors?
      Just keep going as if nothing's wrong?
      Without the sensor information the AP has absolutely no idea how fast it's going, or what to do to keep the aircraft in the air.
      I guess it could estimate the speed based on the thrust generated by the engines, but even then the pilot should decide what to do.

    10. Re:300 million miles by 0123456 · · Score: 1

      So what do you propose the aircraft does when it receives invalid input from the wind speed sensors?

      The expected response from the pilot is to fly a predetermined pitch and power program which keeps the plane at a safe speed. So why not just have the computer raise some alarms and do that automatically?

    11. Re:300 million miles by tibit · · Score: 1

      Finally someone who gets it!!! Thank you. Finally. I thought the comments will end and I'll never see someone who has a clue. Thank you again! We need more of you.

      --
      A successful API design takes a mixture of software design and pedagogy.
    12. Re:300 million miles by Anonymous Coward · · Score: 0

      AF447, where they flew into the supercells that everyone else in the area was avoiding (and that autopilot doesn't route around)? Oh right, after that screwup (the one that precipitated the idiocy that followed) there was plenty of human error and the real issue was that neither human was left 'in control.'

      Chain of command isn't right for most things, but it is appropriate on planes. Had the pilot on the right not been a panicked fool and pulled on the stick to keep the plane in stall, the other pilot (who was doing the right thing, pushing the stick forward to adjust the AoA and gain speed and control) would've calmly descended to a safer altitude where the plane was easier to manage, thus buying enough time for either the autopilot or the experienced captain to come back and assume control.

      Airbus's choice of asynchronous control schemes bears some of the blame as this crash would not have happened (at least by the failure modes that caused it) in a Boeing or other synchronous stick system.

      TL;DR: There was massive human error of AF447, the greatest part was that 2 pilots fought for control without realizing it, causing indecision that lead to the crash. Autopilot might've helped in some portions of the error chain but the original root cause was something that was not avoidable with autopilot.

    13. Re:300 million miles by Anonymous Coward · · Score: 0

      You can't use a self-driving Kia Soul! If you do, you'll have to re-index everything to the life expectancies of hamsters!

  5. The fear of lack of control. by jellomizer · · Score: 1

    The biggest issue with self driving cars is the same issue why people don't like flying, or Cloud Based solutions.

    They are taking something and putting their trust to someone or something else.

    The only comfort is numbers saying if it is indeed safer or not.

    --
    If something is so important that you feel the need to post it on the internet... It probably isn't that important.
    1. Re:The fear of lack of control. by Altanar · · Score: 4, Insightful

      As far as I'm concerned, letting humans drive is putting trust in the other human drivers around me, and frankly, I don't trust them at all. I'd feel much safer if manual driving was illegal.

    2. Re:The fear of lack of control. by Impy+the+Impiuos+Imp · · Score: 1

      Sci-fi and futurists have already dealt with this argument -- when robot drivers are much safer, people will start to question whether it's ethical to let humans drive.

      See also robot surgeons.

      --
      (-1: Post disagrees with my already-settled worldview) is not a valid mod option.
    3. Re:The fear of lack of control. by CastrTroy · · Score: 1

      You're right, everyone else should be in a driverless car, except me.

      --

      Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
    4. Re:The fear of lack of control. by Impy+the+Impiuos+Imp · · Score: 1

      By the way, watch for this: Expensive cars and bleeding edge tech fans will have it first, just as happened with nav radios. This tech will then move to midline then low-end cars. Soon, only old beaters driven by poor will be without.

      Along will come a pontificating politician with a government plan to equip these old cars with safer robot drivers. People will applaud.

      --
      (-1: Post disagrees with my already-settled worldview) is not a valid mod option.
    5. Re:The fear of lack of control. by Anonymous Coward · · Score: 0

      [Fixed]
      As far as I'm concerned, letting OTHER humans drive is putting trust in the other human drivers around me, and frankly, I don't trust them at all. I'd feel much safer if I drove all the cars.
      [/Fixed]

      I can't even get my GPS to put me on the fastest route. you can pry my steering wheel from my cold dead hand.

    6. Re:The fear of lack of control. by lister+king+of+smeg · · Score: 1

      just no custom "slightly faster roms" or Beta testing on the freeway

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    7. Re:The fear of lack of control. by FatAlb3rt · · Score: 1

      Pretty sure that the classics won't be required to be retro-fitted a la no seatbelt requirement if it wasn't originally stocked with seatbelts.

    8. Re:The fear of lack of control. by CanHasDIY · · Score: 1

      As far as I'm concerned, letting humans drive is putting trust in the other human drivers around me, and frankly, I don't trust them at all. I'd feel much safer if manual driving was illegal.

      Important part bolded. Let's see it again, just for emphasis:

      I'd feel much safer if manual driving was illegal.

      Hmm... why is that eerily familiar sounding phrase making me break out in cold sweats?

      I'd feel much safer

      Scary.

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
  6. Numbers, The Law, Reality of Attention by Dripdry · · Score: 1

    so... let's do some math here: they'd need to build 1,000 self-driving cars and drive them each 300,000 miles. Or, 10,000 cars for 30,000 each, etc etc

    Essentially, the DMV is saying that unless Google installs this in a huge fleet of vehicles then it's a no go. Still, it's a start.

    Also, basically the laws say that since someone has to be behind the wheel and ready to take control, what is the person going to do? Sit there for an hour worried that the car is going to fail? That's awful, I'd never in a million years do that.
    The whole reason driving is enjoyable is that one can sort of reach a "flow state" while doing it. It keeps you engaged and aware for long periods. Merely sitting there waiting for software to fail on my 15 hour car trip would see me turn into a madman, frankly.

    --
    -
    1. Re:Numbers, The Law, Reality of Attention by h4rr4r · · Score: 5, Insightful

      Driving is enjoyable?
      Since when?

      Sure a race track is enjoyable, twisty deserted roads can be fun, but 99% of driving is mind numbing boredom.

    2. Re:Numbers, The Law, Reality of Attention by NoNonAlphaCharsHere · · Score: 1

      Airline pilots seem to be able to do it without going insane, although, admittedly, they don't need the reaction/response time that a driver does. If Something Bad (TM) happens to the plane's autopilot, you've got (up to) minutes to recover, in a car, possibly (down to) tenths of a second.

    3. Re:Numbers, The Law, Reality of Attention by jbrandv · · Score: 2

      Tell this to Capt. Sully. When you are travelling over 200 MPH the closing speed is so fast you can't really react fast enough. Um, tell it to the Geese too!
      An airline pilots job is hours of boring flights punctuated by moments of shear terror.

    4. Re:Numbers, The Law, Reality of Attention by h4rr4r · · Score: 4, Informative

      Heck, modern planes even try to fix the problems itself. In the famous case of Colgan Air 3407(crashed near Buffalo NY) after shaking the yoke to alert the pilot the autopilot attempted to trade altitude for speed to get out of a stall. The human pilot overrode this safety feature and killed everyone on board by attempting to gain altitude and thus turned a recoverable stall into a crash.

    5. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 2, Funny

      punctuated by moments of being deathly afraid of scissors?

    6. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 1

      So you hate to drive. That makes you a perfect candidate for a driverless car - go on with your bad self. Some of us live in places where the roads aren't mind-numbingly boring, and drive cars - sports cars, muscle cars, hot rods, rat rods, kustoms, exotics, etc. - which make it fun to drive. Your experience != everyone else's.

    7. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      Please pay attention to what you're reading.

      First off, the 300 million mile figure is from a researcher at Stanford, not from the DMV, so the DMV isn't saying that at all.

      Second, the new law currently requires a driver ready to take control, because the self-driving cars are considered unproven. I'd expect that once self-driving cars have sufficiently proven themselves (according to whatever criteria the DMV decides), they'll be allowed on the roads without a person ready to take the wheel.

      To put it another way, this is an intermediate step before self-driving cars are fully approved. No, it's not useful as a consumer item under these rules, but that's not what these rules are intended for -- they're intended to allow Google (and others) to put their self-driving cars into actual road conditions so they can continue to develop and improve them.

    8. Re:Numbers, The Law, Reality of Attention by h4rr4r · · Score: 1

      To be clear muscle cars are boring as shit to drive. Might as well just put it on rails and have a fast train, not like they can turn anyway.

      I love driving on twisty abandoned roads, or rally or on race tracks. Public streets however are not the place for that. That kind of driving is not safe, and that is the whole point.

    9. Re:Numbers, The Law, Reality of Attention by NoNonAlphaCharsHere · · Score: 1

      Sully had minutes, otherwise they all would have died. And this new "radar" thing extends that 200mph closing distance something wonderful. I'll give you the geese problem. Still, the "chances taken" percentage inside two-second reaction times is WAY higher for drivers than for pilots.

    10. Re:Numbers, The Law, Reality of Attention by Cro+Magnon · · Score: 1

      Driving is enjoyable?
      Since when?

      Sure a race track is enjoyable, twisty deserted roads can be fun, but 99% of driving is mind numbing boredom.

      Often punctuated by 1% sheer terror!

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    11. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 1

      To be clear muscle cars are boring as shit to drive. Might as well just put it on rails and have a fast train, not like they can turn anyway.

      I love driving on twisty abandoned roads, or rally or on race tracks. Public streets however are not the place for that. That kind of driving is not safe, and that is the whole point.

      On the contrary. Rally cars are boring as shit to drive. Might as well just put it on rails and have a roller coaster, not like it can accelerate anyway.

      I love driving on long straightaways, especially fresh paved ones - public streets, country highways. Driving is perfectly safe if one follows speed limits, and that is the whole point.

      See what I did there, Mr. Subjective?

    12. Re:Numbers, The Law, Reality of Attention by 0123456 · · Score: 1

      Airline pilots seem to be able to do it without going insane

      That's because they almost never have the autopilot fail and dump the problem on them.

      And when that does happen, about half the time they seem to crash and burn (or sink, in the AF447 case). Which shouldn't be surprising, given that a situation so bad the autopilot can't handle it is almost certainly so bad that a pilot who has barely touched the controls since their last simulator training and has spend the last six hours fantasizing about the blond in seat A3 can't handle it.

    13. Re:Numbers, The Law, Reality of Attention by 0123456 · · Score: 1

      On the contrary. Rally cars are boring as shit to drive. Might as well just put it on rails and have a roller coaster, not like it can accelerate anyway.

      Yeah, 0-62mph on gravel in 2.3 seconds and being able to go round corners at insane speeds sucks ass.

      See what I did there, Mr. Subjective?

      Made a silly and blatantly false comparison?

    14. Re:Numbers, The Law, Reality of Attention by h4rr4r · · Score: 1

      I see that you made a ridiculous claim.

      If you enjoy muscle cars you might as well just ride in one. You enjoy riding in a fast moving object, not driving.
      If you follow the speed limit you might as well drive long straight stretches in an econobox, it will go 65mph as well as a muscle car.

    15. Re:Numbers, The Law, Reality of Attention by pod · · Score: 1

      The point is you're assuming what people like and don't like, based purely on your own experience, which you deem to be "normal" and "typical". This is a common attitude. I don't drive very often, and for me driving, even on city streets during rush hour, is enjoyable.

      This is all beside the point. Driving is now a candidate for automation, regardless of its perceived enjoyability. Lots of things that many people used to enjoy have been automated or otherwise went the way of the dodo. It's just not a factor we should be considering, and there's no benefit to arguing about it. If you enjoy driving, you'll just have to find something else to enjoy while your car is driving itself in the future.

      --
      "Hot lesbian witches! It's fucking genius!"
    16. Re:Numbers, The Law, Reality of Attention by pod · · Score: 1

      I mean, this is not bad, losing the ability to drive even if you like to to it. Think of all the time you spend driving, and what you might be able to do with it now that you have it available again. You can do work, read a book, watch a movie, study, learn something new, sleep. Now that you don't need to see out the windows, you can have heavy tinting, you can be nside the car naked, masturbating, having sex, having an orgy even! There will be a whole range of things we haven't thought of yet that we'll be doing while in private transit. Surely, you will be able to find something enjoyable to do.

      --
      "Hot lesbian witches! It's fucking genius!"
    17. Re:Numbers, The Law, Reality of Attention by gorzek · · Score: 1

      If you want to drive for pleasure, do it on your own property or a private track.

      Public roads are not there for you to get your jollies on--indeed, doing so makes you a menace to everyone else using the roads for their intended purpose.

    18. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      I'll counter that with the Air France flight from Brazil. Computer crashed the plane into the ocean and killed everyone on board.

    19. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 1

      Driving as a necessity can indeed be boring in the way that you describe. I do drive for fun. I drive on racetracks, I drive on twisty deserted (mountain) roads, I drive in traffic. I drive on feeder streets. I stop at stop lights. I maintain my lane. It is fun. All of it. I do try to stay out of rush hour though but only because most of those people have a distinct need to be somewhere and I do not. I like to be fair when I use the roads.

      If driving is not fun (stoplights and all), then get a car that can elicit that excitement from you. Driving is about the most ultimate activity I can think of. While not quite as good as sex, it lasts a lot longer. ;)

    20. Re:Numbers, The Law, Reality of Attention by h4rr4r · · Score: 0

      You mean AF447?
      Because that was really the humans failing when the computer through its hands up at the problem.

    21. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      You must be driving the wrong car and in the wrong place. Driving on the public road can be fun but I never found it any bit enjoyable in the US

    22. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      People who say they like driving:
      A) Live in the country where your other choice is to walk 25 miles
      B) Have never lived in an even suburban area
      C) Don't actually own a car or do much driving.

      There is nothing enjoyable about driving. The fun fades during your first year. Unless you're a new driver, it's not enjoyable even in a sports car. You have dipshit cops hiding behind hills and trees out to get you and no matter how good you are at dodging them, it will eventually happen.

      You have complete retards like you and I driving with you. The best driver is a complete retard when 1 more car enters the equation because you don't control them. Anyone who lives in a populated area, especially the suburbs, where the ONLY way to get around is to drive every insignificant distance, should know very well aware of how boring, mundane, frustrating, tedious, and time wasting driving is. And expensive. If not, they're lobotomized.

      It's not even necessarily dangerous, it's just a soul sapping horrible chore if you live near other people.

    23. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      Maybe you should get a motorcycle.

    24. Re:Numbers, The Law, Reality of Attention by tibit · · Score: 1

      The autopilot didn't fail. The system is designed to disengage various autopilots as sensor data becomes dubious. That's the only sane behavior of such a system. The pilot was no in the loop, was caught off guard, and failed, like they almost all do in such circumstances. Humans seem to be broken that way, at least at the present time.

      --
      A successful API design takes a mixture of software design and pedagogy.
    25. Re:Numbers, The Law, Reality of Attention by Anonymous Coward · · Score: 0

      You literally, without a hint of irony or self-awareness, claim that you know more about what a stranger enjoys than they do, and THEY'RE the one making a ridiculous claim?

      You are an idiot, a liar, or both. Which is it? No other options can exist, and any claim on your part to the contrary is an irretractable confession that it's "both".

  7. this is how the end begins by atarione · · Score: 1

    clearly this is how computers / machines will rise against us...

    self driving cars in CA will become ENRAGED by the clueless jackasses you have to deal with driving here.. and will rise up and destroy humanity (doubtless they will enlist the computers of people who don't watch enough cat videos as allies, computers seem to love cat videos).

    I for one welcome our new self driving car overloads.

    yes this is how the world ends... Self Driving Car ROAD RAGE.. right before you killed by machines remember I called it.

    --
    actually I am happy to see you, however that is in fact a banana in my pocket.
  8. I wanna "Ask Slashdot" on this by OzPeter · · Score: 3, Interesting

    I have been thinking about driverless cars and I'd love to ask the people at Google (or where ever) how they cope with several real life issues
     
    * Emergency vehicles in general
    * Vehicles on the side of the road. In general you move over to the other side (road,next lane etc) to give them some room. But where I am (VA) its an offense if you fail to move over when passing a cop car on the side of the road.
    * Temporary speed limits posted during road works
    * School zones
    * Really bad weather where you can't even see 20 feet ahed of you
    * Looking down the road and predicting that there will be an issue and doing your best to avoid it (ie slowing down/lane changing to avoid the person on the phone who is weaving from side to side)
    * Crap lying all over the road (saw lots of rocks on a mountain road yesterday)
     
    I'm sure there are lots of other "interesting" situations that human drivers have to deal with day to day that would be difficult to encode into hueristics for the self driving cars.

    --
    I am Slashdot. Are you Slashdot as well?
    1. Re:I wanna "Ask Slashdot" on this by compro01 · · Score: 3, Informative

      * Temporary speed limits posted during road works
      * School zones
      * Really bad weather where you can't even see 20 feet ahead of you

      Given that speed limit signs are fairly standardized and well-defined, having the system recognize them and act appropriately shouldn't be an insurmountable problem.

      As for the weather, self-driving cars will have much more flexible sensing than the Mk1 eyeball. Fog, etc. is considerably more transparent to IR and radar than it is to visible light.

      --
      upon the advice of my lawyer, i have no sig at this time
    2. Re:I wanna "Ask Slashdot" on this by BasilBrush · · Score: 2

      I don't know why you think any of those are difficult situations.

      * Emergency vehicles in general

      They're just other vehicles - they might be doing unusual things, but any auto-driver system has to allow for the fact that any vehicle may do unusual things. They are only limited by the laws of physics not the rules of the road. And it's easy to detect flashing blue lights and sirens and give priority.

      * Vehicles on the side of the road. In general you move over to the other side (road,next lane etc) to give them some room. But where I am (VA) its an offense if you fail to move over when passing a cop car on the side of the road.

      Stationary vehicles are the very simplest vehicles to avoid.

      Temporary speed limits posted during road works

      The technology for vision systems to interpret road signs is already there. Google already reads speed limit and restricted turn signs from their street-view cars to add to their ground-truth database for navigation.

      * School zones

      Signs as above. And schools don't tend to move so will be in the navigation database anyway.

      Really bad weather where you can't even see 20 feet ahed of you

      Auto-drivers aren't limited to the visible light spectrum as human drivers are, and are thus potentially able to see better in poor visibility scenarios. And of course would have more discipline to adjust the speed appropriately to the weather and visibility than the average human.

      * Looking down the road and predicting that there will be an issue and doing your best to avoid it (ie slowing down/lane changing to avoid the person on the phone who is weaving from side to side)

      People expect other drivers to be sensible and obey the rules of the road far too often. I'd expect an auto-driver to always allow for the possibility that another car may do something unpredictable. And if and when they do so, the auto-drivers reaction time would be quicker than a human.

      * Crap lying all over the road (saw lots of rocks on a mountain road yesterday)

      If a human can see an obstacle, an auto-driver will be able to see it.

      I'm not saying all these things actually exist in an auto-driver yet. But they are all achievable with current technology. And I'd expect them to all be there in any system that actually passed the test-plan suggested.

    3. Re:I wanna "Ask Slashdot" on this by Ichijo · · Score: 1

      Given that speed limit signs are fairly standardized and well-defined, having the system recognize them and act appropriately shouldn't be an insurmountable problem.

      Except that the posted speed limit only applies in ideal conditions. You can be driving below the posted speed limit yet still be ticketed for driving at an unsafe speed.

      So the self-driving vehicle would, at all times, need to choose a reasonable and prudent speed that is equal to or less than the posted speed limit.

      And I wonder how often the human behind the wheel will get impatient with the computer for driving so slowly and take the wheel, or how often a following motorist will get impatient and honk or whip around, or how often rear end collisions will occur as a result of following motorists driving too fast for conditions.

      --
      Any sufficiently unpopular but cohesive argument is indistinguishable from trolling.
    4. Re:I wanna "Ask Slashdot" on this by MavenW · · Score: 1

      As I understand it, this is exactly what Google's strategy is--rack up millions of miles in the car, under as many different situations as they can. Record EVERYTHING. If there is ever a situation where the car chooses a wrong strategy and the driver has to take over, flag it for review. Evaluate the situations and see if the engineers think the car did it right. If not, FIX THE ALGORITHM.

      Over time, the car should be handling more and more obscure situations. The collective experience of all the cars in the fleet can be funneled into the master algorithm, and it shouldn't be long before each car will have the equivalent of millions of miles of driving experience. And mistakes where the car chooses badly in a new obscure situation should only happen once. After which the algorithm gets updated and the fleet software gets a revision.

      The computer algorithm may not have a great ability to come up with the correct decision in a new experience, but it has superb rule-using abilities. They just need to add rules (human picked rules) every time they see a new obscure situation. Since there are an infinite number of possible new obscure situations, it will never reach absolute perfection. However, it will approach perfection relatively quickly. Especially after widespread adoption. They could get a lot more experience miles if commercial and civilian drivers submitted instrument recordings of all miles driven. More especially every time they thought the car made a mistake. Most especially if it resulted in an accident.

      Besides, it doesn't need to be perfect--it only needs to be substantially better than human drivers. The bar is set quite low.

      I expect all the situations you have brought up have been encountered by Google's cars more than once. If all of these situations have not been handled in the algorithm by now, I would be quite surprised.

    5. Re:I wanna "Ask Slashdot" on this by FatAlb3rt · · Score: 1

      I think the point the OP was getting to with emergency vehicles is that you are required to move to the right. I wouldn't think it would be terribly difficult for emergency lights to be detected in a similar manner as school zone signs.

    6. Re:I wanna "Ask Slashdot" on this by hedley · · Score: 1

      Scenario: stoplights out for 2 wide 4lane boulevards with multiple left turn lanes.

      Cop is in the centre directing traffic, alternately waving some through etc. This scene is bad enough when there are humans running the show. What would
      GOOG's car do? That would be my Q. Give it a couple of tests, need to make a left and a need to go straight through. (add some bad weather for good
      measure (thus the reason for the lights out).

      H.

    7. Re:I wanna "Ask Slashdot" on this by Anonymous Coward · · Score: 0

      Given that speed limit signs are fairly standardized and well-defined, having the system recognize them and act appropriately shouldn't be an insurmountable problem.
       
      This could be one area where I could see the question of a car being hacked as valid. What does the car do if a vandal changes a speed limit sign? Suddenly a stretch of 65 mph road dips to 5 mph.

    8. Re:I wanna "Ask Slashdot" on this by Anonymous Coward · · Score: 0

      You seem to be under the impression that self-driving cars drive largely blind, relying heavily on a priori knowledge like maps data. While they can and do certainly take advantage of that, the cars are loaded with sensors that allow it to drive much the same way humans do, by looking at the surroundings and acting appropriately. So they will see (or hear) emergency vehicles and pull over. They will see vehicles on the side of the road and move over. They will see temporary speed limits, school zone signs, etc, and respond accordingly. They will recognize poor visibility (although they are likely to have superior sensors to humans in such conditions, and may not need to slow down as much.) Etc.

    9. Re:I wanna "Ask Slashdot" on this by Nimey · · Score: 1

      That may depend on where you live. Here the minimum speed on a highway is 45 MPH and max varies from 55 to 70. As long as you stay between those values you don't get ticketed for speed.

      --
      Hail Eris, full of mischief...

      E pluribus sanguinem
    10. Re:I wanna "Ask Slashdot" on this by Nimey · · Score: 1

      It could fall back to checking its internal map and saying "yup, map says this is a highway and there's no scheduled construction, so there's no way that says 5mph. I'll go with what the map says".

      --
      Hail Eris, full of mischief...

      E pluribus sanguinem
    11. Re:I wanna "Ask Slashdot" on this by quantaman · · Score: 1

      Given that speed limit signs are fairly standardized and well-defined, having the system recognize them and act appropriately shouldn't be an insurmountable problem.

      Except that the posted speed limit only applies in ideal conditions. You can be driving below the posted speed limit yet still be ticketed for driving at an unsafe speed.

      So the self-driving vehicle would, at all times, need to choose a reasonable and prudent speed that is equal to or less than the posted speed limit.

      And I wonder how often the human behind the wheel will get impatient with the computer for driving so slowly and take the wheel, or how often a following motorist will get impatient and honk or whip around, or how often rear end collisions will occur as a result of following motorists driving too fast for conditions.

      That might actually be a fairly significant issue.

      People tend to be better at waiting when they're actively doing something (hence all the walking in the airport), I'm guessing a lot of the tendency to speed is due to the fact that drivers generally aren't doing much, and if the computer is in charge they'll be doing even less. Will people just lay back and accept the computer driver, or will they get missed off at doing the limit and just sitting there, so they'll take over so they can go faster.

      --
      I stole this Sig
    12. Re:I wanna "Ask Slashdot" on this by tibit · · Score: 1

      The reasonable and prudent speed for the self driving care doesn't equal one for a human driver. Those things will probably go speed limit most of the time if car-to-car communications happen.

      --
      A successful API design takes a mixture of software design and pedagogy.
    13. Re:I wanna "Ask Slashdot" on this by Joe_Dragon · · Score: 1

      and trying to do that work zone 45 when other cars are flying by at 55-70 can be very unsafe.

      Now maybe work zone speeds need to be only when workers are present and maybe loser when they are walled off as they are on most big work zones on major highways.

    14. Re:I wanna "Ask Slashdot" on this by BLKMGK · · Score: 1

      Emergency vehicles won't be a problem, they already emit signals - generally using light - to allow red lights to change in their favor. Having a self driving car detect these same signals and properly react should be no problem. Likewise the moving over for a police car situation, it will detect it and take action as needed. As busy as VA roads are - think 66 - moving over for every cop car on the side of the road isn't always possible anyway. BTDT...

      --
      Build it, Drive it, Improve it! Hybridz.org
  9. The Land of Fruits and Nuts by stevegee58 · · Score: 0

    My preconceived notions regarding California have been further reinforced.

    1. Re:The Land of Fruits and Nuts by MyLongNickName · · Score: 1

      Why? Because they are taking measured steps toward a possibly better future but able to pull the plug if it presents a public harm?

      --
      See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
    2. Re:The Land of Fruits and Nuts by stevegee58 · · Score: 1

      Just let me know which roads they're on so I can avoid them. I have my hands full dodging crazy humans without dealing with our cyberdriver overlords to boot.

    3. Re:The Land of Fruits and Nuts by BasilBrush · · Score: 4, Interesting

      Why do Luddites chose to come to Slashdot?

    4. Re:The Land of Fruits and Nuts by MyLongNickName · · Score: 1

      Thanks. I figured you were just a clueless individual, but wanted to give you the benefit of the doubt. You confirmed my suspicions.

      --
      See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
    5. Re:The Land of Fruits and Nuts by stevegee58 · · Score: 1

      Oh, I have a clue. It's just I'll probably be dead by the time driverless cars truly come into their own. In the meantime I'd rather avoid them.

    6. Re:The Land of Fruits and Nuts by Anonymous Coward · · Score: 0

      A land filled with Fruits and Nuts is much better than one filled with idiotic anti-technology asshats such as yourself.

    7. Re:The Land of Fruits and Nuts by Big+Hairy+Ian · · Score: 1

      Why do Luddites chose to come to Slashdot?

      If they come anywhere it'll be here where the meat is :)

      /ripley

      --

      Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.

  10. "Prepare for crash" code ... by perpenso · · Score: 2

    ... does what its programming tells it to do, avoid hitting other vehicles ...

    Its a bit of an assumption to believe that the driving software has that single goal. Staying on the road seems to be something the software is already considering. I wouldn't be surprised if existing software already has "prepare for crash" code that tightens seat belts, unlocks doors, ... maybe even sends an "oh shit" text message to the road side assistance service.

    1. Re:"Prepare for crash" code ... by h4rr4r · · Score: 1

      Most cars already have some of this. Hit the brakes too hard and the seatbelts tighten. I would imagine you are correct and this will only get better and better.

      Some sort of inflato hans-device would be awesome. Of course I always wanted a 5 point harness in my cars.

    2. Re:"Prepare for crash" code ... by o5770 · · Score: 1

      Most cars already have some of this. Hit the brakes too hard and the seatbelts tighten.

      That's not some magical AI, it's mechanical function.

    3. Re:"Prepare for crash" code ... by h4rr4r · · Score: 2

      Yes, and your point is?

      This is something we do without AI, with AI we could do a lot more. Maybe even have airbags on the outside of the car, to set off right before a crash.

    4. Re:"Prepare for crash" code ... by gorzek · · Score: 1

      We don't need "AI" to do any of the things a car does, just a system that reads various sensors and reacts to them as it is programmed to do.

    5. Re:"Prepare for crash" code ... by thesandtiger · · Score: 1

      Most modern cars have sensors that detect a crash in progress and pull those things off before the kinetic energy is a risk to human occupants. It happens incredibly fast, and only when you get hit etc.

      I would rather that, to be honest, than something that anticipated what it imagined to be an inevitable crash before contact was made just on the off chance that the crash wasn't actually inevitable.

      What I do see happening though is an automated driving car (or just any car that has automatic control of braking and speed) to try and adjust speed ahead of a potential crash as a way to avoid or minimize it. For a fully automated car, I would expect it to be constantly scanning the area and detect something approaching in a dangerous fashion in plenty of time to avoid a collision entirely and not even have to swerve - it would simply adjust speed to leave a hole for the fast moving interloper much as a good human driver will give plenty of space when they see an erratic driver.

      --
      Since I can't tell them apart, I treat all ACs as the same person.
    6. Re:"Prepare for crash" code ... by flimflammer · · Score: 1

      Were you trying to take that point somewhere?

    7. Re:"Prepare for crash" code ... by h4rr4r · · Score: 1

      How is what you are describing not AI?
      Hell, how is that not intelligence?

      My sensors(eyes) show my brain your post and I make this response because my upbringing as programmed me to be like this.

    8. Re:"Prepare for crash" code ... by gorzek · · Score: 1

      It's not "AI" as most people would think of it. That's just rather typical programming: wait for inputs, analyze them, produce outputs (responses.)

      What about that makes it more "AI" than any other interactive software?

    9. Re:"Prepare for crash" code ... by h4rr4r · · Score: 1

      How is " wait for inputs, analyze them, produce outputs" not exactly what all intelligence does?

      Serious question.

    10. Re:"Prepare for crash" code ... by Shrike82 · · Score: 1

      It's not "AI" as most people would think of it. That's just rather typical programming: wait for inputs, analyze them, produce outputs (responses.)

      Artificial Intelligence. The ability to react to stimuli and make decisions based on those stimuli is a solid definition of intelligence. Yes, there are others. The fact that it's programmed intelligence makes it Artificial. Trust me, it is AI and the argument over semantics is moot and unproductive.

      --
      You can advertise in this sig from as little as £99.99 a month!
    11. Re:"Prepare for crash" code ... by gorzek · · Score: 1

      Then virtually all software is AI and people can stop asking "when will we have AI?" :-p

    12. Re:"Prepare for crash" code ... by Shrike82 · · Score: 1

      A lot of software that we have today can be considered as AI, but like everything there's various degrees. Siri is a form of AI, albeit a crappy one. Wolfram Alpha can perform basic reasoning. Asimo (that little creepy robot) does some stuff that even 30 years ago would have been considered science fiction. We have AI now, just not the Skynet or Sentient Intelligence that people expect from books and films.

      --
      You can advertise in this sig from as little as £99.99 a month!
  11. If Microsoft made cars... by Aerowin · · Score: 1

    I'm sure you all remember the jokes about if Microsoft made cars (http://www.hcs.harvard.edu/pnw/microsoftjoke.htm). Yes this is Google, but they still have the computer mentality in them.

    1. Re:If Microsoft made cars... by Anonymous Coward · · Score: 1

      If Microsoft made cars.
      (a href="http://www.hcs.harvard.edu/pnw/microsoftjoke.htm")If Microsoft made cars.(/a)
      Replace the ( ) with < >

    2. Re:If Microsoft made cars... by compro01 · · Score: 1

      Google isn't making the car (The test vehicle is a Prius, IIRC) , just the driver.

      --
      upon the advice of my lawyer, i have no sig at this time
  12. Not me. by Anonymous Coward · · Score: 0

    If I had a choice between my wife and a self driving car, the self driving car would be my first choice - you have to be there.

    Since I hate to drive and half the time I'm sleep deprived, my first choice between me and a self driving car would be the self driving car.

    I can only see an upside here.

    Actually there is one downside - speed limits. My Garmin gets a lot of speed limits wrong and if Google is as bad, I see a LOT of speeding tickets - especially in South Carolina's 25MPH "Commercial Zones" in their shitty little towns.

    1. Re:Not me. by Cro+Magnon · · Score: 1

      If I had a choice between my wife and a self driving car, the self driving car would be my first choice - you have to be there.

      Since I hate to drive and half the time I'm sleep deprived, my first choice between me and a self driving car would be the self driving car.

      I can only see an upside here.

      Actually there is one downside - speed limits. My Garmin gets a lot of speed limits wrong and if Google is as bad, I see a LOT of speeding tickets - especially in South Carolina's 25MPH "Commercial Zones" in their shitty little towns.

      How well would it update speed limits? On one of the roads I take, there's road work, and the normal 55 MPH is now 45. I could see getting a ticket there, especially in a "work zone" (even though I didn't see anyone actually working there :p ).

      --
      Slow down, cowboy! It has been 4 hours since you last posted. You must wait another few hours.
    2. Re:Not me. by Anonymous Coward · · Score: 0

      Why don't you just let your wife drive?

    3. Re:Not me. by FatAlb3rt · · Score: 1

      Speed zone changes have to be uploaded to a central database. I'd bet that self-driving cars are going to be online all the time. You give it your destination, it maps the route, and picks up info like speed zones, construction, traffic volume, and weather & road conditions.

    4. Re:Not me. by jgtg32a · · Score: 1

      Wife by definition means she is a woman.

  13. Ahhh The Future Belongs To The Machines by Zamphatta · · Score: 4, Funny

    So if I come out of the grocery store and my car's not there, it might not be stolen?

  14. Self driving cars are nice, but... by Anonymous Coward · · Score: 0

    ...honestly the only place I would use them is on expansive highways.
    I unfortunately live in Texas, with relatives/family all over the place.
    The drive between major cities is long and horribly boring, half the time I'm tempted to just put it in cruise control and smash something on the wheel to keep it going straight.

  15. 725,000 ***representative*** miles ... by perpenso · · Score: 1

    725,000 representative miles

    I hope by "representative" they mean diverse traffic and road conditions that represents the various things that a driver will experience over their driving career.

    While on this thought I don't expect self-driving vehicles to be universally permitted. There will probably be limitations on the conditions under which an automated driving mode may be used.

    As a side note, I believe the FAA adopted a statistical approach to safety regarding civilian space flight. That space flight should be no more hazardous than early commercial airline travel, I think the 1920s was used a baseline. Well at least that was the tentative plan at some point according to a presentation from some folks up at Mohave Spaceport.

    1. Re:725,000 ***representative*** miles ... by bluefoxlucid · · Score: 2

      There's a lot of weirdness here.

      For example, 725k miles for any incident, but if we look at only "fatal" crashes it skyrockets to 300M? There's a disconnect here: if we look at only "fatal" crashes, I'm pretty sure we can smash up Google cars every 30k without killing anyone and make it to 300M.

      If you say, "well it has to be 1 in 300M because that's how often a fatal crash occurs and we want to reduce fatal crashes," you're talking about something completely different. 1 crash in 300M miles isn't likely to be 100% fatal; beating 1 in 725k miles reduces the number of actual incidents, which reduces the number of incidents with a likelihood of being fatal, which theoretically should reduce fatal crashes.

      Further, if the car handles better than a human, it may reduce damage and injury in crashes. That means if it crashes more often than a human, it may still cause less total injury and less economic costs--a $300 scuffed bumper rather than a $3000 smashed in side panel, a gashed and mashed-in door on a sideswipe rather than a car's front end smashing straight through and crippling the driver on a T-bone, plain old reduced impact speeds due to better braking and faster brake reaction, etc.

      I'm sure Mercedes-Benz, Volvo, and Google can all work something out. Benz has those pre-crash systems (best in the industry); Volvo has reactive braking (alerts the driver, pre-charges the brakes, but doesn't apply braking force until a crash is unavoidable; in a self-driving car, it'd react by suggesting braking to avoid collision to the driving program, which in theory should acknowledge that a likely collision is about to occur and apply brakes along with all other proper maneuvering). If the car is reasonably capable of driving itself, there's no reason it can't reduce incidence and reduce damage in incidents with all these collision prediction systems and automatic pre-crash systems that can brake, steer, and do all kinds of other shit for you.

    2. Re:725,000 ***representative*** miles ... by FatAlb3rt · · Score: 1

      Fully agreed. I'd like to see a statistician pop in here and talk about confidence levels too. Just because the average is 1 in 300M for a fatal crash, that doesn't mean that by driving 300M miles with no crash means that we're safer. That's like flipping a coin where tails = fatality. If you flip it twice and both come up heads, that just means you got lucky.

    3. Re:725,000 ***representative*** miles ... by perpenso · · Score: 1

      Fully agreed. I'd like to see a statistician pop in here and talk about confidence levels too. Just because the average is 1 in 300M for a fatal crash, that doesn't mean that by driving 300M miles with no crash means that we're safer. That's like flipping a coin where tails = fatality. If you flip it twice and both come up heads, that just means you got lucky.

      I am not a statistician but I've had enough stats classes to know that there should be a "all other things equal" somewhere in your statement. :-)

    4. Re:725,000 ***representative*** miles ... by Neil+Boekend · · Score: 1

      What the article says is "When there are 300M miles without a fatal accident then we can be 99% certain the self-driving car is better than the average human". This means there is a 1 % chance the car just was lucky all the time.

      It's like flipping a coin 7 times, getting heads all the time and saying there is a 99% chance the coin is rigged (2 heads for example), since the chance of that with a fair coin is only ~ 1%. It could be chance, but most probably it's rigged.

      No I am no statistician. I just used my HAVO (high school) maths.

      --
      Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
  16. Note for Mods by MyLongNickName · · Score: 0

    Troll does not mean that you disagree with an opinion. I disagree with the OP, but he brings up logical points that you can at least counter. Mods who downvote posts like the above should have their mod rights taken away.

    --
    See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
  17. So what I want to wonder is... by Anonymous Coward · · Score: 0

    How long do you think it will take for this law to be overturned?

  18. Finally! Better driving! by bdemchak · · Score: 1

    Finally! California has figured out a way to improve the driving on California roads. Can an occasional system crash be any worse than what we see every day?? ... and this way, texting addicts can indulge safely. Sure. :)

  19. Promising but obstacles ahead by Anonymous Coward · · Score: 0

    This has great promise not only for safety but for fuel efficiency and road capacity. I don't know how useful it is to be able to switch to manual control when the car in front of you is only inches away from your bumper and you're doing 90 mph.

    But there are going to be lots of problems from water and ice to dust and snow. How to keep the probes clean?

    Also, there's bound to be the practical jokers and Asimov-style murders where the perfectly functioning probes are spoofed into fatal crashes without leaving a trace for the investigators.

  20. make them mandatory asap by thetoadwarrior · · Score: 1

    The sooner we can stop self-entitled douche bags from ruining the roads the better.

  21. If driverless cars are safer ... by Anonymous Coward · · Score: 1

    then anyone who choses to drive and has an accident is wilfully negligent. Discuss.

    If driverless cars are safer, the only logical conclusion is to ban humans from driving.

    Afterall, who could possibly oppose making the roads safer?

    1. Re:If driverless cars are safer ... by Anonymous Coward · · Score: 0

      You're assuming everyone can afford to buy a new car.

  22. What's the point.. by planckscale · · Score: 1

    ...if a driver needs to be behind the wheel? I mean yeah it's great and all you don't need to put your hands and feet anywhere but if you're supposed to be alert watching that the car doesn't make a mistake then what's the difference? You still can't text, read the paper, play cards, eat dinner, whatever - or can you?

    --
    Namaste
    1. Re:What's the point.. by ColdSam · · Score: 1

      Not dying in a car crash?

  23. Don't let the Law keep ya down by deburg · · Score: 1

    Personally, I am pleased that a goverment is pro-actively taking account of new technologies and laying out a road-map.

    Until recently, in my country, electric cars were not allowed to being driven on the roads (except on private land), because there was a law requiring each car to have paid road tax. That same law also stated that road tax were calculated based on engine displacement , which is not present on an electric car (duh). As such, until the law was ammended in 2010 (and set to take effect in 2011), the only ECars were golf carts and demo-cars that were ferried on trucks.

  24. A better test by gr8_phk · · Score: 1

    Just let them do whatever they want but don't provide any exemption of liability. When they are prepared to bet the company in lawsuits, then the cars is probably safe enough. Just remember, when 2 of them crash, there is not question who caused the accident/damage/death. When the company is willing to accept that responsibility I'd give them a shot.

    And BTW, the reason this is easier to do today is because brake-by-wire, steer-by-wire, radar systems, etc have already been developed by the auto industry. We've had systems that can drive at highway speeds on winding dirt roads for years. Google did not lead the way - though they are now sharing the leading edge and pushing for legalization.

    1. Re:A better test by h4rr4r · · Score: 1

      No problem, they can do that now.
      Just declare bankruptcy and sell the company off for pennies on the dollar to $NOT_CURRENT_NAME.

      For a nice example look at what GM did recently.

    2. Re:A better test by BasilBrush · · Score: 1

      Liability is easily managed by using insurance. And insurance companies are excellent at gauging risk. If an auto-car system passes the test for being statistically safer than a human driver, then for sure there will be insurance companies willing to take on that liability.

    3. Re:A better test by CanHasDIY · · Score: 1

      Liability is easily managed by using insurance. And insurance companies are excellent at gauging risk. If an auto-car system passes the test for being statistically safer than a human driver, then for sure there will be insurance companies willing to take on that liability.

      Presuming that one of these things causes an accident, who do you think would be held liable for damages?

      --
      An enigma, wrapped in a riddle, shrouded in bacon and cheese
    4. Re:A better test by Anonymous Coward · · Score: 0

      Why not apply a similar standard of driver fault to that we give humans? It works especially well because everything can be recorded.

  25. Miitgating associated risk of terrorism by Vincent+Van+Googol · · Score: 1

    What is the best way (short of a panopticon) to minimize the inevitable delivery of remotely controlled explosive devices to a target location via one or more driverless (and passengerless) vehicles? I'm assuming here that the bad guys would find a way around initial methods used to ensure a human passenger was present in the vehicle.

    --
    "My God...it's full of spam"
    1. Re:Miitgating associated risk of terrorism by Bramlet+Abercrombie · · Score: 1

      The terrorists have already been doing this, but nobody wants to talk about it.

  26. Legal Liability? by Bramlet+Abercrombie · · Score: 1

    Who is legally responsible if a driverless car causes property damage or bodily injury? Will insurance companies cover these cars?

  27. What's the point? by Drunkulus · · Score: 1

    Why is this being developed? If saving lives were the issue, we could start on that with simple driver training. Germany requires a week of driver ed. in order to get a license, and although they have many roadways without a speed limit, they have half the rate of fatalities that the US does.

  28. They will fail and here's why by ackthpt · · Score: 1
    • They don't tailgate - you're not a real California driver unless you've driving 1 mm off the bumber of the car ahead of you.
    • They don't drive at unsafe speeds for road conditions - speeding & tailgating in the rain is a California speciality
    • They won't pass in the bike lane or on shoulder of the road - if you don't like our driving, stay off the sidewalk
    • They won't stop and hold up traffic, despite having the right of way, to wave through someone who clearly doesn't have it or is waiting for them to get through the stop so they can go, because they don't realize the other driver thinks they're being "nice"
    • They won't go the speed of traffic, which is usually 15 MPH faster than the speed limit, thus causing a lot of accidents as other drivers try to get around them
    --

    A feeling of having made the same mistake before: Deja Foobar
  29. Exemptions by ThatsNotPudding · · Score: 1

    Californina and Florida have added a retroactive exemption for large Cadillacs driven only by wizened knuckles and the tops of bouffant hairdos.

  30. Got nothing on Slashdot by Anonymous Coward · · Score: 0

    We've had self-posting bots for years.

  31. Why the 99% confidence interval? by Dr_Barnowl · · Score: 3, Interesting

    The usual standard for a statistical "proof" is held to be a 95% confidence, or a p value of 0.05 that the hypothesis is wrong.

    Using a 99% confidence interval is skewing the numbers away from the usually accepted standard of proof, which makes me suspicious about the motives of the person proposing it.

    1. Re:Why the 99% confidence interval? by camperdave · · Score: 1

      Well, if you're going to err, this is one of those times when you want to err on the side of caution. I have no problem with a 99% confidence interval over a 95% one, because people's lives are at stake.

      --
      When our name is on the back of your car, we're behind you all the way!
    2. Re:Why the 99% confidence interval? by Daniel+Dvorkin · · Score: 1

      Ah, the p-value less than or equal to alpha=0.05, the Holy Grail of Statistics 101 students and the bane of working statisticians.

      It's not a bad standard for most uses, because "a one in twenty chance that you're wrong" corresponds pretty well to most people's idea of acceptable risk most of the time. But there's nothing magical about it, and in applications relating to human safety, it's quite common to use a more conservative standard, e.g. alpha=0.01 for safety and alpha=0.05 for efficacy in clinical trials. IOW, we're willing to accept a higher risk that the drug in question is essentially a placebo than that the drug will actually kill patients. A similar standard applies, seems to me, when it comes to turning autonomous and potentially lethal machines loose without human control.

      --
      The correlation between ignorance of statistics and using "correlation is not causation" as an argument is close to 1.
    3. Re:Why the 99% confidence interval? by kasperd · · Score: 1

      Well, if you're going to err, this is one of those times when you want to err on the side of caution.

      But what does side of caution even mean in this case? What if you instead of 99% confidence only had 80% confidence? It would mean that there is at least 80% chance that an automatic car is safer, so in other words is there at most 20% chance that a manual car is safer. How does the principle to err on the side of caution lead you to choose the approach, which has less than 20% chance of being the safest of the two?

      --

      Do you care about the security of your wireless mouse?
    4. Re:Why the 99% confidence interval? by Anonymous Coward · · Score: 0

      The usual standard for a statistical "proof" is held to be a 95% confidence, or a p value of 0.05 that the hypothesis is wrong.

      Using a 99% confidence interval is skewing the numbers away from the usually accepted standard of proof, which makes me suspicious about the motives of the person proposing it.

      We're not talking about testing a hypothesis that will then have to be validated by follow up studies. We're talking about determining the safety of a product. If you were only 95% certain that a food product were safe, would you eat it? If your doctor is 95% sure that a drug is not going to kill you, should he prescribe it to you? Product safety should meet far higher standards than scientific hypotheses in a single study.

    5. Re:Why the 99% confidence interval? by Anonymous Coward · · Score: 0

      I suspect this article may have had input from a google employee for the stats stuff. it's not like you frequently find degree level stats in journalistic articles.

      I recon 99% was used because if you use a 95% interval it sounds too much like "there is a 5% chance this thing will crash!".

      Also, the 95% interval has probably already been driven past, so they wouldnt be able to make a blog post about how safe they have just proven their car to be.

    6. Re:Why the 99% confidence interval? by Anonymous Coward · · Score: 0

      95% confidence is a convention, but no statistician anywhere ever would call it grounds for statistical proof. Even when used properly it describes a one in twenty chance of coincidence, which is pretty high when compared to the connotation of the word "proof". Moreover, it's often used improperly (multiple comparisons etc).

    7. Re:Why the 99% confidence interval? by Anonymous Coward · · Score: 0

      That's not how confidence intervals work. 80% confidence means that under repeated sampling one would expect the procedure used to give the correct result 80% of the time, meaning that in each experiment there's a 20% chance of falsely demonstrating that the machines are safer.

      Bayesian analysis can be used to make statements like yours, but not traditional confidence intervals.

    8. Re:Why the 99% confidence interval? by camperdave · · Score: 1

      If the goal is to pick the higher safety rate, and there is a 99% confidence that the automatic car has the higher safety rate, wouldn't that be better than only being 95% confident that the automatic car has the higher safety rate? Suppose I had two decks of well shuffled cards, and I dealt each deck into two piles, A and B. With the red deck there are 51 cards in pile A and only one card in pile B. With the blue deck, I've got 40 cards in pile A and twelve cards in pile B. If your job was to pick out which pile had the ace of spades, which deck would you rather play with, the red or the blue?

      --
      When our name is on the back of your car, we're behind you all the way!
    9. Re:Why the 99% confidence interval? by kasperd · · Score: 1

      I guess you are right, but in that case I'd argue that they have been answering the wrong question. What we really want to know is whether automatic cars or manual cars are safer. And the answer they can provide says, we don't know.

      --

      Do you care about the security of your wireless mouse?
    10. Re:Why the 99% confidence interval? by kasperd · · Score: 1

      If the goal is to pick the higher safety rate, and there is a 99% confidence that the automatic car has the higher safety rate, wouldn't that be better than only being 95% confident that the automatic car has the higher safety rate?

      Nobody is saying that 95% is better than 99%. If there is 95% certainty that automatic cars are safer, would you rather be using manual or automatic cars? Is anybody going to say that they would rather use a manual car because there is only 95% certainty, that the automatic car is safer?

      As for the 99% we'll get there over time. But even before there is enough data to reach the 99% you still need to make decisions on what to do until then.

      --

      Do you care about the security of your wireless mouse?
    11. Re:Why the 99% confidence interval? by tibit · · Score: 1

      That would mean that 5% of published research comes up with entirely wrong conclusions, with everyone around nodding approvingly. The "usual standard" is up to the researcher to determine. If one thinks that p=0.05 is sufficient in all cases, or somehow "usual", one only contributes to the problem.

      --
      A successful API design takes a mixture of software design and pedagogy.
    12. Re:Why the 99% confidence interval? by tibit · · Score: 1

      As another anecdote, in functional safety engineering, safety integrity level 4 (SIL 4), the highest one, calls for the probability of dangerous failure per hour to be between 10^-9 and 10^-8. Alternatively, if a safety function is called upon only once a year or rarer (think an emergency stop, process shutdown, panic button, etc), the probability of failure on such a demand should be between 10^-5 and 10^-4.

      If you'd apply it to a car, say think of your drive-by-wire car's ignition shutdown button used in an emergency situation while driving (expected to be used less than once a year, duh). It would be expected to fail, or not shut down the engine, at most for one in 10,000 drivers who called upon it. I don't know if automotive people use IEC 61508 SIL approach, but that's the automotive analogy.

      --
      A successful API design takes a mixture of software design and pedagogy.
    13. Re:Why the 99% confidence interval? by Thatmushroom · · Score: 1

      A 95% confidence interval is the usual convention, but there's nothing sacred about that number. In my view, it's usually more important to consider expected loss, where you multiply the probability of something going wrong by how much damage would occur for each possible outcome. Let's take a walk and see why you might want a much tighter CI for something like this:

      At some point, an automated car will be involved in an accident that causes a fatality. It might be the other car's fault, it might be equipment failure, it might be the AI's fault, but it will happen. People are not very good at understanding risk or teasing out causality, so there will be an outcry calling for heads at Google to roll for this. People will see a new process that takes away control from them, freak out, think that they would do better with their own hands on the steering wheel, and there will be a major push to ban driverless cars. States like California might not give in, since their economy is closely tied into tech companies, but red states? Fuggedaboutit. Emotions trump statistics, and the longer we have to adjust to the idea of driverless cars, the less likely it is that there will be a strong resistance. At this point, we're considering a 5% chance that we were wrong and that driverless cars aren't demonstrably better than humans behind the wheel. A 5% probability of driverless cars being just as unsafe as manually driven cars, multiplied by the lives lost because of delays in implementation after the general public freaks out, is a pretty big expected loss. Truth be told, this probably understates the expected loss, because we're shooting to prove that driverless cars are safer, but it will take a lot of time before we know how much safer. We've been driving for a century, our country is permeated with the car culture, and we have a reasonable grasp of the risks involved (from a statistical, not emotional, level). It will take a while to accumulate the same kind of data on driverless cars.

      This also considers a driverless car surrounded by cars with human drivers in the state of California. It's important to keep the scope of the testing in context. It's probably true that we can extend this to the roads in various conditions with varying compositions of driver+driverless cars, but it's easy to see how many people will have reservations about that. Heck, look at this thread.

      I can't wait for driverless cars to roll out en masse, not least because of how many lives they will save, but it really is the cultural and legal side that will hold things up across the country. If we have to wait a little longer for the rollout to happen because we're looking for a higher standard of proof, that might be an optimal course of action once you consider that we're dealing with the general public.

      Yes, IAAStatistician.

      --
      You zap the moderators with a wand of humor! The moderators resist!
    14. Re:Why the 99% confidence interval? by camperdave · · Score: 1

      Ah! I see your point.

      The real issue isn't going to be what is acceptable to you and I, or to mathematicians and statisticians, but what is acceptable to the general public. The more nines you tack on to the result, the easier it will be for the general public to accept.

      --
      When our name is on the back of your car, we're behind you all the way!
    15. Re:Why the 99% confidence interval? by Anonymous Coward · · Score: 0

      Air bags are only at 30%, and seatbelts something like 50% and the public accepts them for the most part. If auto-drive ever reaches 95% it will have long since been required by the government.

    16. Re:Why the 99% confidence interval? by Neil+Boekend · · Score: 1

      Of course they don't know. Of course they aren't certain. If anyone is certain about anything then they haven't thought it trough.
      The best you can do is give a certainty factor. Most people don't: they even say they are absolutely certain, while that is impossible. Please realize that nobody is ever certain about anything (that's a fact with 99.99999% certainty). We get by with uncertainties. Many people don't even realize that they don't know anything for certain.

      Are you certain that we are not a figment of your imagination? Are you certain the universe is not a figment of your imagination? Are you certain we are not a simulation of a system in a 83 dimensional universe made by beings who oversimplified their universe simulation? Can you ever be certain? With that uncertain basis on which you base the world, can you ever be certain of anything? If the rest of the universe is just a figment of your imagination then how can the car be safe or even exist?

      If that is to much: They say that if the car made 300M miles without a fatal accident then there is 99% chance the car is better than an average driver and 1% chance it was lucky all that time. If they said it was 100% certain it was better then I'd be calling them fools.
      However: 99% certain is enough in most cases, IMHO including this one.

      --
      Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
    17. Re:Why the 99% confidence interval? by kasperd · · Score: 1

      Are you certain we are not a simulation of a system in a 83 dimensional universe made by beings who oversimplified their universe simulation?

      I can say with 90% certainty that you got the number of dimensions wrong. If you left out the number of dimensions, your theory would sound plausible. Look at the quantummechanical duality between particles and waveforms. It looks like a clever hack to simulate continuous physical phenomena with a quantized representation in the simulation. Can anybody come up with a more plausible explanation for why that duality exists? I am just hoping I'll be the one to discover the vulnerability that will allow me to break out of the sandbox. It would give me access to all knowledge in the universe and unlimited computational power.

      --

      Do you care about the security of your wireless mouse?
  32. Robots ho! Whar me robot hos? by Impy+the+Impiuos+Imp · · Score: 1

    So human drivers have about one death or fatal accident per 300 million miles? That seems awfully high and does not jive with another stat I heard years ago (which, I admit, I have never seen anywhere else) that road deaths statistically shave 6 years off the average lifespan. The latter implies something like 20 million miles lifetime, which also seems an order of magnitude too high.

    --
    (-1: Post disagrees with my already-settled worldview) is not a valid mod option.
  33. I vote YES by stanlyb · · Score: 1

    After seeing today a pretty serious crash at the intersection just next to the school, and seeing that obviously the parents of the kid were not with seat-belts, and being in pretty bad condition, but THANK GOD, the kid was in child booster, and with seat belt, and was feeling PERFECTLY well (surprise, surprise), and seeing how one woman who just stopped while waiting for the green light, just next to the cop, and (surprise, surprise) was WITHOUT seat-belt, and the cop did nothing, and she for some strange reason decided to put the seat-belt on.............
    I are you starting the feel the fear that i felt bringing my kid to the school???
    Do you know how many times the drivers signaled when they changed the lane, or left/right turns? I have to admit, a saw a few cases, obviously not from California..
    So, just to summarize it, anything, i repeat ANYTHING is better than the regular Californian driver. Period.

    1. Re:I vote YES by Anonymous Coward · · Score: 0

      I dare you to make less sense.

  34. if that's the question by Chirs · · Score: 1

    then you should only engage the autopilot when you're sleep deprived, upset at my wife and in a hurry to get home.

    I agree with the grandparent...any self-driving car I buy should be better than me on a typical drive for it to make sense. I want it to drive on the highway for 9 hrs and let me sleep or read a book or watch a movie or play with my kids.

    1. Re:if that's the question by h4rr4r · · Score: 1

      No, because when you're sleep deprived, in a hurry and upset you will forget to turn it on.

      It really only needs to be better than the average driver, since odds are most of us are pretty near the average. Distribution curves, how do they work?

    2. Re:if that's the question by tsmithnj · · Score: 1

      so basically i would have it on 24x7.

    3. Re:if that's the question by Anonymous Coward · · Score: 2, Interesting

      The problem is that most people overestimate their own driving skill - a 1981 study (Svenson) found that 93% of American drivers tested ranked themselves as being above the median in driving ability.

    4. Re:if that's the question by SolitaryMan · · Score: 4, Informative

      I never get upset at your wife.

      --
      May Peace Prevail On Earth
    5. Re:if that's the question by h4rr4r · · Score: 1

      That's good, bad sign to get upset at people who do not exist.

    6. Re:if that's the question by MrNaz · · Score: 1

      Why would I be upset at your wife?

      --
      I hate printers.
    7. Re:if that's the question by farble1670 · · Score: 1

      I want it to drive on the highway for 9 hrs and let me sleep or read a book or watch a movie or play with my kids.

      i want it to play with my kids while i sleep or read a book.

    8. Re:if that's the question by WolfWithoutAClause · · Score: 1

      Tell that to any religious person.

      --

      -WolfWithoutAClause

      "Gravity is only a theory, not a fact!"
  35. robot cars = sprawl enablers by doom · · Score: 4, Insightful
    I don't think there's any question that automated cars can beat human beings at safety, nor is there any question that they can reduce pollution just by driving more evenly (not to mention by drafting each other, "tailgating" to form car-trains).

    The trouble with them is that they'll take the sting out of long commutes. You already have people who think it's a good idea to spend four hours a day driving for the sake of cheaper real estate. What if they up it to six hours a day when they don't have to stare at the road?

    Note: cutting a problem (pollution, car-deaths) would do no good if you double the miles.

    1. Re:robot cars = sprawl enablers by Anonymous Coward · · Score: 0

      Quite the opposite, actually.
      You argue that cutting [presumably by half] the pollution or car-deaths per mile would do no good if you double the miles driven.
      While in this scenario, no decrease in pollution or car-deaths would result, you would in fact do plenty of good.
      You'd be opening up the opportunity for people to take advantage of cheaper real estate without any increase in pollution or car-deaths. Many people would argue that this is in fact a good thing.
      Conversely, we can look at your claim and take it to its logical conclusion: it would be great if we stopped driving entirely. No pollution, no car-deaths, everyone's happy.
      You're forgetting the other side of the equation: transportation has value.

    2. Re:robot cars = sprawl enablers by rwv · · Score: 1

      Note: cutting a problem (pollution, car-deaths) would do no good if you double the miles.

      Doubling miles would double pollution... but your point about unintended consequences is a good one.

      On the flip side, when sprawl happens businesses open up in the sprawl. Given the cost of square footage in developed "urban" and "suburban" areas an expanding area of civilized sprawl might make middle class living more affordable for more people (which has its own advantages). Myself... I pay a buttload of money on a condo mortgage and live 7 miles from work. I recently (last month) realized that running to work was feasible... so I've been doing a bit of that for a fitness benefit. Aside from access to more convenient commute options, the short distance between house and work doesn't provide much other benefits.

    3. Re:robot cars = sprawl enablers by HeckRuler · · Score: 1

      The wonderful thing about them is that they'll take the sting out of long commutes.

      Imagine what those thosands of engineers will do with their gained hour of productivity?
      ...Mostly angry birds or equivlent. But in addition to everyone being a little happier with their bleak existence when they show up to work or at home, some will actually do meaningful work instead today's defacto STARE AT ROAD.

    4. Re:robot cars = sprawl enablers by Beryllium+Sphere(tm) · · Score: 1

      The commuters would still be paying the substantial expenses of car operation.

    5. Re:robot cars = sprawl enablers by Anonymous Coward · · Score: 0

      Europe has shown this problem is easy to solve. Tax gasoline heavily and people will tend to conserve it, by getting more efficient cars and avoiding driving. Hybrids and electric cars make the equation more complex, but raise the price enough and people will act. Problem is taxes are looked down upon by voters, so proposing new ones tends to be DOA.

    6. Re:robot cars = sprawl enablers by Anonymous Coward · · Score: 0

      I doubt it would work out that way. I have a commute that long, mostly by train and it's pushing the boundaries of comfort. There are a number of people who have a commute as long as you're suggesting but they have access to such comforts as tables and toilets.

    7. Re:robot cars = sprawl enablers by Anonymous Coward · · Score: 0

      Imagine this: Your electric car is waiting for you as you leave your front door, drives you to work, drops you off and cruises to a nearby recharge station where it has reserved a spot. You message it as you leave your desk and it's out front waiting for you when you get to the curb - fully charged, ready to take you home, drop you at your front door and park in your garage.

      Possible benefits:

      reduced air pollution - shifted to energy generation plants

      reduced noise pollution

      traffic reduced as people aren't looking for parking spots

      reduced crashes

      reduced insurance premiums?

      reduced stress!

    8. Re:robot cars = sprawl enablers by Anonymous Coward · · Score: 0

      how about electric engines, which is available NOW, and that will solve your little pollution problem with or without drafting each other

    9. Re:robot cars = sprawl enablers by doom · · Score: 1

      how about electric engines, which is available NOW, and that will solve your little pollution problem with or without drafting each other Actually, of course, electric cars don't solve pollution problems, they just transfer them to wherever the power plants are. If the power is relatively clean (e.g. nuclear) than this is a win, if the power is dirty, i.e. like the half of our power that's still generated by coal, you'd be better off burning gasoline.

      I wish the delusion that electric==clean was only confined to slashdot trolls like this, but it's clear that it exists in the wider populace as well. The electric buses they use in San Francisco proudly display "Zero Emissions" decals...

  36. I want an Auto-PIT maneuver! by trout007 · · Score: 1

    If someone cuts me off I want my AI driver to perform a PIT maneuver and make them slide off the cliff.

    --
    I love Jesus, except for his foreign policy.
  37. reduce DWIs by Wansu · · Score: 1

    The robot is your designated driver.

    --
    Wansu, th' chinese sailor
  38. What about animals? by Anonymous Coward · · Score: 0

    Anyone know what would be considered if an animal was crossing the road. Lets assume good weather and street conditions, no traffic, etc. Driving out in the middle of the country when an animal is on the road.

    There are so many different situations, that I'm not sure how a computer can make a decision.
    What if:
    - A turtle is trying to cross road. Human drives around if safely. (what would computer do?)
    - A deer standing/staring at you (blinded by the headlights). Sometimes it'll flee, sometimes it just stands there. Something as big as a deer or cow, you'd want to try to avoid. Been a few times where I've swerved to miss it (deer bolts from left to right across road, swerve left to avoid the impact.)
    - A small animal (possibly R.O.U.S.) on the road. Well, if it's safe, a human would go around, if it wasn't safe to go around or brake fast enough, then the small furry animal is getting smoked. What if the small furry animal was a person? Human would then decide (usually) to not smoke said mammal and do something more dangerous.

    Overall,
    How would these computers drive in the country?

    1. Re:What about animals? by 0123456 · · Score: 1

      There are so many different situations, that I'm not sure how a computer can make a decision.

      The computer will shut down and give control back to the driver, who will smash into the animal before they even realise that's why the computer shut down. And that will go down on the accident records as 'human error'.

  39. economic impact? by mspring · · Score: 1

    If there's no need for drivers anymore, where will those people who used to be drivers make their living of?

    1. Re:economic impact? by 0123456 · · Score: 1

      If there's no need for drivers anymore, where will those people who used to be drivers make their living of?

      Rich people will hire them to prove they're superior to the rest of us because they have real drivers instead of computers.

  40. Demographics? by gninnor · · Score: 2

    I wonder if the aging population will end up pushing this into reality. We will not make mass transit is not going to work on a large enough scale, and for many transportation needs are only met by POVs. It will become yet another device to assist people's independence, and that I believe will push the technology and laws as the need for it increases.

    1. Re:Demographics? by Anonymous Coward · · Score: 0

      Nothing designed to enhance independence does - they only increase your dependence and controllability if anything should happen to them, even if they work fine. Most people are boned without a car, or phone or whatever. All of those things are easily used to further control you. Bills, licenses, registration fees, taxes, tracking, licenses, records, etc.

  41. Re:Robots ho! Whar me robot hos? by 0123456 · · Score: 1

    That seems awfully high and does not jive with another stat I heard years ago (which, I admit, I have never seen anywhere else) that road deaths statistically shave 6 years off the average lifespan.

    A quick back of an envelope calculation shows that would require nearly 500,000 road deaths every year in America. So, yes, that statistic seems completely bogus, which is probably why you've never seen it anywhere else.

  42. It's inevitable.... by erp_consultant · · Score: 1

    Today's drivers have a dizzying array of distractions. Between cellphones and increasingly video game-like car controls it's a wonder there are not more accidents. I, for one, think that touch screen controls in cars are a really bad idea. It forces you to take your eyes off the road to adjust the radio or the air conditioning or other items that with manual controls can be operated by feel. I have an older car and all my controls are manual. I know exactly where the radio buttons are without having to look at them. When I travel and rent cars I notice that they all have this same sort of design. Everything is operated by a touch screen and you have to drill down through menu options to get what you want. It looks cool but it's distracting to operate.

    Cellphones, of course, are self inflicted distraction. From behind the signs of obvious. Slow swerve towards the next lane followed by the head jolting up and an abrupt lane correction. When I see idiots like that I stay far, far away. Secretly I hope that one of these morons smashes their car into a tree (without injuring anyone else of course) so that I can pull over behind them, get out of my car and laugh uncontrollably at their stupidity. But I digress....

    If we can prove that robot driven cars are safe then I think it's a good thing...if only to save us from ourselves. Todays cars are more powerful, more distracting and there are simply more of them on the road.

    1. Re:It's inevitable.... by Anonymous Coward · · Score: 0

      I, for one, think that touch screen controls in cars are a really bad idea. It forces you to take your eyes off the road to adjust the radio or the air conditioning or other items that with manual controls can be operated by feel. I have an older car and all my controls are manual. I know exactly where the radio buttons are without having to look at them.

      You know, when radios started being popular with cars, people were saying the exact same things about them.
      Quite frankly, I've seen many a person who can operate a touch screen without looking at it as well, so your arguement of "I know where the buttons are" isn't that great either.

    2. Re:It's inevitable.... by tibit · · Score: 1

      I couldn't agree more. Touch-screens for cabin controls are a man-machine interface disaster from hell. People who design those things should be whacked on the head until it gets to them that it's a bad idea.

      --
      A successful API design takes a mixture of software design and pedagogy.
    3. Re:It's inevitable.... by erp_consultant · · Score: 1

      I think it's a perfect example of technology attempting to solve a problem that doesn't exist. The old school dials and push buttons work. They never break. I can operate them without having to take my eyes off the road. Now voice activated controls are a different matter. Those I see a lot of value in but the hunt and peck touch screens are just a disaster waiting to happen. I bet they cost a lot of money to repair as well.

    4. Re:It's inevitable.... by tibit · · Score: 1

      The labor involved is same, parts cost similarly as well. In any case you either replace the entire module, or replace the entire board subassembly that has all the controls on it, or the screen+touchscreen assembly. So that's not the issue.

      The issue is that designers cave in to some perceived need for "hot and trendy" solutions that simply don't apply in given problem domain. Touchscreens for tablets and perhaps even phones: great idea. Touchscreens for body controls in a car: silliness.

      --
      A successful API design takes a mixture of software design and pedagogy.
  43. Do we have real AI? by EnergyScholar · · Score: 1

    But since we have not mastered true AI yet

    Are you sure about that? Your understanding of the AI basis for self driving cars may be flawed. Google's AI (not exclusive to Google) is based on 'artificial life' that was 'encouraged' to evolve intelligence more than 20 years ago. It probably qualifies as both 'strong AI' and 'friendly AI'. Computers are the just medium it uses to communicate with us and are not its basis technology. It's called Quantum Neural Network technology (no useful wikipedia entry yet exists), and its basis is non-abelian anyons interacting in a 2DEG. See the complete scientific works of Stuart Kauffman, Steven Wolfram, Robert Laughlin, and David Deutsch for details.The reason you can't read about this new technology is because the original project that developed it was classified, so all derivative technologies (such as Google's self driving cars) are also classified, and all people who work on this technology must sign ironclad NDAs.

    1. Re:Do we have real AI? by tibit · · Score: 1

      Huh?

      --
      A successful API design takes a mixture of software design and pedagogy.
  44. We should probably accept 50%, honestly... by Sanians · · Score: 1

    We should probably be willing to accept 50%. We're dealing with a driver who remembers every detail of an accident perfectly, and whose decisions can be analyzed, and the driver upgraded to never make the same mistake again. Even if there's a small increase in accidents initially, the accident rate will likely go down so quickly that more lives will be saved by using the cars sooner than would be saved if we waited until we were certain that these cars are safer than humans.

  45. Ok by Anonymous Coward · · Score: 1

    Sounds cool..Who is at fault in an accident? The car? :)

    1. Re:Ok by kasperd · · Score: 1

      Who is at fault in an accident? The car?

      Faulty cars can cause accidents, even if they are driven manually. If you could save 10 out of every 100 people killed in traffic, wouldn't you want to do it? Or do you really rather want a 100 people killed, where you can put the blame on some person, than have 90 people killed where you cannot find a person to blame?

      Or put differently, if you had 90 people getting killed in accidents, would you really want to sacrifice another 10 lives just so you can put the blame for the first 90 on somebody?

      My point is, the ability to put the blame on somebody is not important. What is important is the total amount of accidents.

      --

      Do you care about the security of your wireless mouse?
  46. Turn your phone off and DRIVE! by Anonymous Coward · · Score: 0

    Today's drivers have a dizzying array of distractions.
    Self-inflicted.
    The number of cars and motorcyclists weaving in and out
    of lanes in the Bay Area has gotten to the point where you are nuts to be
    driving with these sorts of things enabled.

  47. why all the resistance? by Anonymous Coward · · Score: 0

    This is the solution the elderly need so that they can maintain their freedom without risking the safety of others.

  48. Bright Future by Anonymous Coward · · Score: 0

    Computer Controlled Cars offer such a huge advantage over human driven cars, that to me it's just a matter of time before they are everywhere. I see already a lot of people are afraid, but that's just because they've already assimilated the risks of human driven cars into their lives already.

    Some of the advantages won't be seen until the number of computer controlled cars hits a critical mass, but imagine this future:

    -Make full use of existing infrastructure. No need to build new bullet trains, instead the computer controlled cars will maximize the use of the existing road system.

    -Reduced congestion. Cars will coordinate with other cars on the freeway to reduce traffic, eliminate accidents, and improve your travel times.

    -Reduce pollution. Computer controlled cars will not need to idle in traffic or at stop lights as often because they can schedule their arrival in advance. Those car commercials where cars weave in and out of each other will happen at every intersection. Additionally, the computer will control acceleration/deceleration for the best fuel consumption.

    -You'll be able to schedule your car to pick up your kids from school. Or drop you off at the airport and then return home.

    -You'll be able to rent a car (eg: Zip Car) from an app on your phone and it shows up and takes you to your destination. Many people won't even need to own their own car.

    -Automatic Valet. Your car will drop you off at the store, then go park itself in a high density parking garage. Finding a space will be no issue since the garage will communicate with the car to assign it a location.

    -Imagine you have a medical emergency, your car could act as your own ambulance and rush you to the hospital. Perhaps even alerting other cars of an emergency along the way.

    -Computer controlled cars will probably also tend to be smaller. Families might have 2 small cars instead of a minivan. Just lock the kids in the second car and take a peaceful ride.

  49. Who Benefits? Nobody here on this board! by Anonymous Coward · · Score: 0

    First implementations will be used for Armed Homeland Security Surface Drones. Elimination of Truck Driver Jobs. Surveillance Drones. Why was there an article planted in the Global MSM about giving 'Autonomous Robots' Human Rights? This is to protect 'The Man's" robot against destruction by the Plebs.

    The USA is a third world country. The Americans living below the Median Income will not be buying these. These people live on food stamps. Who will buy these? What will be their first uses? Who makes them? Do they also make military hardware?

    So can I have a Fat Reefer in one hand and a Glass of Crown Royal in the other? A Glock on the seat, and be safe and invisible behind my ultra tinted glass?

    I mean the 'Perfect Machine' is driving and my implanted chip passes the security check by the Two Tap Drone flying over head and The Enforcement Droid Series 209.

    If all this shit passes remote muster, how now will they justify the 'no cause' 'no warrant' stop and the seizure of my preemo bud and my nine?

    Sure, they 'say' the whole thing is safe. Would you drive a car that a Microsoft Saleman said was 'Safe'. Would you drive a cra that Zuckerberg said was 'private'? Given the shitty state of the art of all the tech shoddiness in the real world, how can anyone buy into this utopian ideal of this Perfect Software? Are Aliens writing it? Is God? This sounds more like utopian silver bullets than fact.

    And really, have the Video Games and iPhones so atrophied the capabilites of today's youngsters that they are too feeble to 'Drive a Car?' You guys don't have the physical strength, the hand eye or the stamina to drive?

    And do you really want to live in a world that complicated, that requires so much on technology? What is the desire to overcomplite?

    How about a walkable city?

    How about the promised cyber commute?

    Where are you going to go in these driverless cars? You don't have any jobs left to go to.

    Who will run the fleet? Halliburtion? Lockheed Marting? Kellog Brown and Root?

    I will not be a first adopter. I will not be endorsing this technology. I am talking to the fricking wall.

    When ED209 is lobbing DU rounds into you mother basement because of that movie you downloaded, I am gonna say I told you so and gloat a big gloat.

  50. The judgement call by YrWrstNtmr · · Score: 1

    The scenario I want to see solved is 'the judgement call'

    Driving along, an oncoming car starts to intrude on your space. So you have to avoid it. There is also a kid chasing a ball across the street. Given the time/distance/speed/angles, you will hit one of the following:
    Other car, almost head on
    Kid
    Ball the kid is chasing
    Tree in the front yard to your right

    Again...you WILL hit something. And, in this case, the least expensive damage to your car is not the right decision to make.

    Humans don't do it right all the time, but something like this must be handled by the logic tree.

    1. Re:The judgement call by geddo · · Score: 1

      I don't understand your point here, are you saying there is a correct answer? which one is it? should the computer driven car assume the life of the child is worth more than the combined lives of both car's passengers? should it consider the lives of its own passengers as having less meaning than that of the child? or should it check the speed at which the two cars are moving v. the distance to the tree v. the distance to the child and make a choice that would have the least impact force? What choice would you make in this situation? What choice would you make if you had your kid in the car? What choice would you make if the oncoming car had a kid in it? What choice would you make if the person in the road was an adult? How many of these do you think you could process in your brain and convert to action in time before the time to make a decision elapsed? I struggle with this idea of machines making decisions v. humans, but humans make selfish decisions all the time that are not necessarily better or worse than the more pragmatic choice a computer may make in this case which I would guess would be to hit an object at the least amount of impact as possible. But of course all this software is written by humans, humans that would be managed and regulated by bureaucracies, so who knows what stupid decisions points would eventually be factored in and dictated by the many lawsuits.

    2. Re:The judgement call by geddo · · Score: 1

      I missed the 'ball' in your options... humans make mistakes ;) I thought you were going for the morality choice here, but I see you intended it to be a question of can it make the correct choice when there is an obvious one to humans. I think there are many ways to accomplish this, object recognition could certainly distinguish a living creature v. an inanimate one and also distinguish the smaller ball vs. the larger tree and fast moving car. Cameras and infrared can be used to distinguish objects along with other sensors and code can be written with logic to make good decisions, I doubt this is going to be the technical equivalent of a bunch of 2 ton Roombas driving around.

  51. Blue Screen by Russ1642 · · Score: 1

    The first time someone is killed by one of these cars the term "blue screen of death" will take on a new meaning.

  52. Lol, no. by Anonymous Coward · · Score: 0

    As expected. Have fun with your terrible idea. Nothing that weighs over 1000 pounds and rolls should ever be unattended by a person, whether controlling it or not. A future where we sit in cars that drive themselves isnt a better future. It's moronic, especially as bad as our computer technology is. Anyone who can't see that does not live in reality.

  53. Suck it, Jetsons! by proca · · Score: 1

    It just goes to show you how you can't possibly predict the future. Although, I think that a flying car would be extremely dangerous if people were allowed to drive them. Are you allowed to tell your car to pick you up from the airport? You would also have tons of person-less cars driving in the airport loop waiting for their overlords.

  54. Imitating/Dealing with A**hole drivers by Anonymous Coward · · Score: 0

    I wonder how this car deals with the human side of driving, or as I call it the A**hole driver symptom... You know, like when you want to make a lane turn, have your indicator on for 5-10 seconds, and the moron in the next lake stays in your blind-spot no matter how you speed up or slow down? My solution is often to floor the accelerator so I can get enough space between us to make a safe change. But this often leads to breaking the speed limit.

    Alternatively, it might need to mimic the moron on the road. Like when a light turns yellow, it should try to beat the light if it can do so safely. If not, people (even crappy drivers) are going to bitch about how much better it would be if they were behind the wheel. I guess is whether it will take calculated risks like most human drivers would (incl. the occasional rule breaking) would influence whether most drivers accept it as a replacement - to mimic their day to day 45-in-a-40-zone activity?

  55. Who's insurance pays when the car has an accident by mrclevesque · · Score: 1

    Who's insurance pays when a driverless car causes material damage or kills someone

  56. What's the point of a self-driving automobile? by Simonetta · · Score: 1

    Pardon me for asking, but what's the point of a robotic self-driving automobile? Simply to show that it can be done?
    Driving is straight forward and most people who buy cars enjoy driving. Maybe in a place that has horrible commutes, one could read a book instead of paying attention to road ragers. But likely not.
    Too drunk to drive? get a taxi. or have someone give you a ride. Or sleep in the bar. We need bars with $15 overnight sleep-it-off bunks more than we need robot cars. Need the car in the morning? Take a taxi back to it.
    Need a cheap chauffeur? Get an illegal alien with a California driver's license.

    1. Re:What's the point of a self-driving automobile? by glitch0 · · Score: 1

      It will be clearly safer than driving manually. So safety is the point.

      --
      -Glitch "We all know Linux is great...it does infinite loops in 5 seconds." - Linus Torvalds
  57. criminal liability and hiding under subcontractor by Joe_Dragon · · Score: 1

    criminal liability and hiding under subcontractor does not work the same way as a civil case. and some one dieing in a auto car crash may end up a criminal case.

  58. won't work in snowy condition by itchybrain · · Score: 1

    Just thought that I would add something I found when I took Sebastian Thrun's AI course.

    The Google self-driving car will not work in snowy condition. This is because the localization algorithm computes the car position using lane lines. So, if there is snow on the ground, lane acquisition will fail.

    The other tidbit I will throw in also is that the Google car favours a right turn over a left turn because it is deemed less risky (i.e., change lane as least times as possible).

    Oh, and one other thing. The Google car does not honk. The Google self-driving car team cannot figure out how to communication effectively with other drivers in this fashion. I

  59. Free licensing for automotive intelligences? by Paul+Fernhout · · Score: 1

    From 10 years ago: http://www.pdfernhout.net/on-funding-digital-public-works.html#what_have_funding_policies_in_automotive_intelligence_wrought

    Consider again the self-driving cars mentioned earlier which now cruise some streets in small numbers. The software "intelligence" doing the driving was primarily developed by public money given to universities, which generally own the copyrights and patents as the contractors. Obviously there are related scientific publications, but in practice these fail to do justice to the complexity of such systems. The truest physical representation of the knowledge learned by such work is the codebase plus email discussions of it (plus what developers carry in their heads).

    We are about to see the emergence of companies licensing that publicly funded software and selling modified versions of such software as proprietary products. There will eventually be hundreds or thousands of paid automotive software engineers working on such software no matter how it is funded, because there will be great value in having such self-driving vehicles given the result of America's horrendous urban planning policies leaving the car as generally the most efficient means of transport in the suburb. The question is, will the results of the work be open for inspection and contribution by the public? Essentially, will those engineers and their employers be "owners" of the software, or will they instead be "stewards" of a larger free and open community development process?

    Open source software is typically eventually of much higher quality ( http://www.fsf.org/software/reliability.html )
    and reliability because more eyes look over the code for problems and more voices contribute to adding innovative solutions. About 35,000 Americans are killed every year in driving fatalities, and hundreds of thousands more are seriously injured. Should the software that keeps people safe on roads, and which has already been created primarily with public funds, not also be kept under continuous public scrutiny?

    Without concerted action, such software will likely be kept proprietary because that will be more profitable sooner to the people who get in early, and will fit into conventional expectations of business as usual. It will likely end up being available for inspection and testing at best to a few government employees under non-disclosure agreements. We are talking about an entire publicly funded infrastructure about to disappear from the public radar screen. There is something deeply wrong here.

    And while it is true many planes like the 757 can fly themselves already for most of their journey, and their software is probably mostly proprietary, the software involved in driving is potentially far more complex as it requires visual recognition of cues in a more complex environment full of many more unpredictable agents operating on much faster timescales. Also, automotive intelligence will touch all of our lives on a daily basis, where as aircraft intelligence can be generally avoided in daily life.

    Decisions on how this public intellectual property related to automotive intelligence will be handled will affect the health and safety of every American and later everyone in any developed country. Either way, the automotive software engineers and their employers will do well financially (for example, one might still buy a Volvo because their software engineers are better and they do more thorough testing of configurations). But which way will the public be better off:
    * totally dependent on proprietary intelligences under the hoods of their cars which they have no way of understanding, or instead
    * with ways to verify what those intelligences do, understand how they operate, and make contributions when they can so such automotive intelligences serve humane purposes better?

    If, for ex

    --
    A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
  60. Consequences. Unintended? by Anonymous Coward · · Score: 0

    Kind of a strange coincidence that right around the time there are announcements for domestic drones (possibly armed), there are State Legalization of "unmanned cars/driverless cars" all over the place. How is this that much different from the FAA allowing for air surveillance (maybe, armed. maybe not) drones?

  61. Fast Lane by LBt1st · · Score: 1

    As long as they stay out of the passing lane I don't care.
    When I'm older, the youngsters will marvel at my ability to operate a motorcycle.

  62. Re:Robots ho! Whar me robot hos? by Neil+Boekend · · Score: 1

    According to a quick google the number of fatal accidents per 100 million miles was 1.11 in 2010. That is with 30,196 fatal crashes and 2,967 billion miles.

    Now this is where my HAVO (Highschool) maths break. Is the 2,72 times less likely to crash equal to 99% certain it's safer?

    --
    Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
  63. blind aren't we by Anonymous Coward · · Score: 0

    We are obviously evolving in a certain order. Electrical cars will replace gasoline, and then it will be much more viable and safer for self driving cars, so much so they will eventually stop looking like cars and more like pods. However that will not replace human controlled transportation as flying vehicles will also be used. So in short, picture electric transportation buzzing around, and flying vehicles streaming over head. How long ? Well with so much conservative thinking going on even by our alleged "intelligent" slashdot posters comprised of IT personnel and such, and naysayers by our bible (first edition & its crappy sequel) and koran thumping oligarchs running our lives, we will not evolve anywhere so quickly.

  64. self driving? by Anonymous Coward · · Score: 0

    terrists would LOVE those cars.

  65. No Personal Jet Packs Or Flying Cars Yet... by Anonymous Coward · · Score: 0

    But Self-Driving Cars Are Here Now -

    http://www.scifiai.com/articles/selfdrivingcars.htm

  66. Please tell me by Big+Hairy+Ian · · Score: 1

    RoboTaxi drivers conversation will have a volume contol!!!

    --

    Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.

  67. No personal jet packs or flying cars yet... by scifiai · · Score: 1

    but self-driving cars are here. My take on it... http://www.scifiai.com/articles/selfdrivingcars.htm

    --
    www.scifiai.com