California Opens Driverless Car Competition With Testing Regulations

← Back to Stories (view on slashdot.org)

California Opens Driverless Car Competition With Testing Regulations

Posted by samzenpus on Thursday May 22, 2014 @07:46AM from the johnny-cab dept.

smaxp (2951795) writes "California just released rules for testing autonomous vehicles on California's roads and highways. Californians will soon be seeing more autonomous vehicles than just those built by the Google X labs. These vehicles offer great promise, such as freeing the driver's attention for productivity or leisure, better safety and less congestion. It will be a while, though, before we see these vehicles on the road. From the article: 'Getting started requires the RMV’s approval of testing under controlled circumstances prior to testing on public roads. The manufactures must insure the vehicles with a $5 million surety bond. Autonomous vehicle manufacturers need a permit and test drivers need a special license. The RMV will receive applications beginning on July 1, 2014, and the permits that are granted will be announced beginning on September 1, 2014.'"

167 comments

Min score:

Reason:

Sort:

The future is now! by Anonymous Coward · 2014-05-22 07:51 · Score: 1

Welcome to the future:
Johnnycab: Please state the street and number.
Douglas Quaid: Drive! drive!
Johnnycab: I'm not familiar with that address. Would you please repeat the destination?
Douglas Quaid: Anywhere just go! Go!
Johnnycab: I'm not familiar with that address. Would you please repeat the destination?
Douglas Quaid: Shit! shit!
Johnnycab: Would you please repeat the destination?
Douglas Quaid: [Quaid rips the Johnnycab out and starts to drive himself] Aaahhh!
I propose a test ... by gstoddart · 2014-05-22 07:52 · Score: 5, Insightful

I call it the aggressive, psychotic driver who makes random, unsafe lane changes, fails to signal, and swoops across several lanes of traffic while doing well over the speed limit.
Lemme see your driverless car handle that, then we'll see.

--
Lost at C:>. Found at C.
1. Re:I propose a test ... by Narcocide · 2014-05-22 08:03 · Score: 5, Insightful
  
  That isn't any sort of a problem for LIDAR at sufficient resolution. It remains to be seen whether it can sufficiently improve traffic flow and accident incidence/mortality rates, but personally I'm more worried about asshats who will purposefully try cloak their cars so that the automated sensors can't even see them at all, just in some misguided attempt to try to prove self-driving cars as unsafe to protect "muh freedomz!"
  Don't get me wrong, I'm personally not interested in one of these self-driving contraptions, but its pretty apparent during any rush hour(s) that at least 90% of the drivers on the road couldn't beat a self-driving car's computer for accuracy to save their own lives on a good day.
2. Re:I propose a test ... by pseudofrog · 2014-05-22 08:03 · Score: 1
  
  Because humans react faster than computers?
3. Re:I propose a test ... by CastrTroy · 2014-05-22 08:05 · Score: 2
  
  Also, let's see how it deals with a bi-directional bike lane where one of the lanes is against traffic. These exist in Montreal, as well as other places I'm sure. Instead of putting 1 bike lane on each side, they put them both on one side of the street. When turning across the bike lane, you have to watch out for cyclists coming from both directions.
  
  Also, let's test in snow. I'm tired of these things that only work in sunny California. It's bad enough that cell phones don't work with gloves on. I would hate it if my car failed to operate when there was a little bit of snow falling.
  
  --
  
  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:I propose a test ... by jcochran · 2014-05-22 08:07 · Score: 1
  
  I call it the aggressive, psychotic driver who makes random, unsafe lane changes, fails to signal, and swoops across several lanes of traffic while doing well over the speed limit.
  Lemme see your driverless car handle that, then we'll see.
  Let's see now.
  Aggressive driver going well over the speed limit cutting in front of me.... Don't really see that as a problem if the autonomous car is doing the speed limit. Yes, there's an interval where the car is too close for safe following, but the aggressive driver fairly rapidly increases the gap. (after all, you did state "going well over the speed limit"). As for failure to signal? I somehow doubt that the programming and sensors for the autonomous car will even notice turn signals. It will however notice the car shifting towards the side of the lane and will likely assume that the car will continue its sideways motion. Given the reaction speed of computer, a lot of the problems caused by aggressive drivers will pretty much go away.
  Random - That's the impression a HUMAN driver would have. A better term for "random" would be "unpredictable". And since the autonomous vehicle would be monitoring the relative location of nearby vehicles, people, and other objects the main criteria is "will that object with its current velocity and potential acceleration impact this vehicle?"
  Unsafe - Just another aspect of your "random" comment. Please see above response.
  Fails to signal - As mentioned, turn signals are not considered a reliable source of data. They are meant as advance warning to us rather slow humans. Stick with the physics based solution.
5. Re:I propose a test ... by Joe_Dragon · 2014-05-22 08:08 · Score: 2
  
  Will it get lost in Death Valley.
  http://www.npr.org/2011/07/26/...
6. Re:I propose a test ... by Mordok-DestroyerOfWo · 2014-05-22 08:09 · Score: 2, Funny
  
  Have you been to California? The Sierras can be ridiculously treacherous. Personally, I'd sooner put my trust in a computer than I would a Texan who's only seen snow on Wikipedia.
  
  --
  "Never let your sense of morals prevent you from doing what is right" - Salvor Hardin
7. Re:I propose a test ... by gstoddart · 2014-05-22 08:10 · Score: 2
  
  No, but humans tend to adapt to things faster.
  Computers ... well, I have less faith in that.
  Not saying it wouldn't out-perform some humans, but the random stupid stuff I see on the road every day tells me they better have a LOT of coverage for corner cases.
  
  --
  Lost at C:>. Found at C.
8. Re:I propose a test ... by Anonymous Coward · 2014-05-22 08:12 · Score: 0
  
  Don't the Google cars encounter this every day on 101?
9. Re:I propose a test ... by Joe_Dragon · 2014-05-22 08:13 · Score: 1
  
  Will they do the speed limit when most of the traffic is going over?
  Will it miss read school zones? Both ways?
10. Re:I propose a test ... by Archangel+Michael · 2014-05-22 08:15 · Score: 1
  
  The real test is to have the driverless cars race 500 miles at a super speedway like Daytona.
  
  --
  Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
11. Re:I propose a test ... by epyT-R · 2014-05-22 08:18 · Score: 2
  
  No, but humans are much more contextually aware than computers.
12. Re:I propose a test ... by Joe_Dragon · 2014-05-22 08:18 · Score: 3, Funny
  
  so they can fail at makeing right turns?
13. Re:I propose a test ... by Ichijo · 2014-05-22 08:19 · Score: 1
  
  When the normal speed of traffic is above the posted speed limit, self-driving cars will drive the speed limit as legally required but will cruise in the right lane as legally required when driving below the normal speed of traffic.
  
  --
  Any sufficiently unpopular but cohesive argument is indistinguishable from trolling.
14. Re:I propose a test ... by Anonymous Coward · 2014-05-22 08:21 · Score: 1
  
  Driverless cars should be required to pass this test on a moonless night, in the rain, on an unmapped road, with GPS unavailable. All those things happen, all the time, and any competent driver is expected to deal with it.
  They should also be required to understand and comply with hand signals and verbal commands given by police officers and construction workers. They should be required to understand gestures from pedestrians.
  Wait, this means a functional driverless car would require A.I.? No shit.
15. Re:I propose a test ... by Joe_Dragon · 2014-05-22 08:22 · Score: 1
  
  so auto drive cars will be slower and put others at risk by going slower then normal speed of traffic
16. Re:I propose a test ... by AK+Marc · 2014-05-22 08:26 · Score: 1
  
  The driverless car, going the speed limit in the presence of the driver you mention who is "speeding", would ignore the aggressive driver. And nothing bad would happen.
  
  Why do so many people complain about those in front of them going faster than them? They are in front of you, and going faster than you. They will not affect you in any way, unless they crash, which happens very rarely (on a per-passenger-mile scale).
  
  --
  Learn to love Alaska
17. Re:I propose a test ... by camperdave · 2014-05-22 08:31 · Score: 1
  
  The real test is to have the driverless cars race 500 miles at a super speedway like Daytona.
  No, the real test would be driving in Cairo, Egypt.
  
  --
  When our name is on the back of your car, we're behind you all the way!
18. Re:I propose a test ... by Anonymous Coward · 2014-05-22 08:35 · Score: 0
  
  They're going slower than the legal maximum limit, which is, well, legal and desirable.
19. Re:I propose a test ... by orasio · 2014-05-22 08:38 · Score: 1
  
  Tell that to google search, or wolfram alpha
20. Re:I propose a test ... by Anonymous Coward · 2014-05-22 08:38 · Score: 0
  
  Around here, you have to watch for cyclists going both ways even if there isn't a bike lane!
21. Re:I propose a test ... by i+kan+reed · 2014-05-22 08:40 · Score: 1
  
  They perform only a little worse than professional drivers in races as far as time goes.
22. Re:I propose a test ... by Ichijo · 2014-05-22 08:41 · Score: 1
  
  No, people who drive above the speed limit put themselves at risk.
  
  --
  Any sufficiently unpopular but cohesive argument is indistinguishable from trolling.
23. Re:I propose a test ... by i+kan+reed · 2014-05-22 08:41 · Score: 1
  
  Your objection is that obeying the law is dangerous.
24. Re:I propose a test ... by AK+Marc · 2014-05-22 08:51 · Score: 1
  
  Random - That's the impression a HUMAN driver would have. A better term for "random" would be "unpredictable". And since the autonomous vehicle would be monitoring the relative location of nearby vehicles, people, and other objects the main criteria is "will that object with its current velocity and potential acceleration impact this vehicle?"
  That's the impression a bad driver would have. The person weaving doesn't say "oh, I traveled 10*RND(x) seconds, I'll change lanes to [left/right]." They say "I wish to be going the speed that should be legally set, and I'll weave to go it. I'll go in the open space to the right." And yes, there are documents indicating that Dallas was notified that they broke State law in setting artificially low speed limits, but they ignored it and set illegally low limits for a while until people started fighting in court, and managed to win. Drivers knew the limits were unenforceable, so many went very fast with impunity, as the limits were illegal.
  
  --
  Learn to love Alaska
25. Re:I propose a test ... by SternisheFan · 2014-05-22 08:52 · Score: 1
  
  ... They should also be required to understand and comply with hand signals and verbal commands given by police officers and construction workers. They should be required to understand gestures from pedestrians.
  Good point. LEO's will need some form of wireless killswitch, or control mechanism on them at all times. Which capability will of course get abused by sociopathic hackers.
  At least the future doesn't look boring...
26. Re:I propose a test ... by Joe_Dragon · 2014-05-22 08:52 · Score: 1
  
  have you even been on I-294?
27. Re:I propose a test ... by SternisheFan · 2014-05-22 08:57 · Score: 1
  
  "Did you ever notice that anyone driving slower than you is 'An Idiot!', and anyone driving faster than you is 'A Maniac!' ???"
  - George Carlin
28. Re:I propose a test ... by dpidcoe · 2014-05-22 08:59 · Score: 1
  
  but personally I'm more worried about asshats who will purposefully try cloak their cars so that the automated sensors can't even see them at all, just in some misguided attempt to try to prove self-driving cars as unsafe to protect "muh freedomz!"
  Screw passive cloaking, lets see how well that self driving sensor package works when getting hit full in the face with some broad spectrum jamming.
  
  Depending on how well driverless cars handle that sort of thing (and how well they can trace any jamming back to the source), I could imagine jamming might replace sawing partway through the tie rods or poking holes in the brake line as the hip new way for arranging accidents for inconvenient people.
29. Re:I propose a test ... by CanHasDIY · 2014-05-22 09:02 · Score: 1
  
  I call it the aggressive, psychotic driver who makes random, unsafe lane changes, fails to signal, and swoops across several lanes of traffic while doing well over the speed limit.
  At night, during a blizzard, on a section of road with more holes than a chunk of Swiss cheese.
  You know, get that realism factor going full-steam.
  
  --
  An enigma, wrapped in a riddle, shrouded in bacon and cheese
30. Re:I propose a test ... by CanHasDIY · 2014-05-22 09:04 · Score: 1
  
  I could imagine jamming might replace sawing partway through the tie rods or poking holes in the brake line as the hip new way for arranging accidents for inconvenient people.
  Yea, because the government totally won't have a backdoor already installed at the factory for dealing with... let's call them "undesirables."
  Nope, nein, no way, huh-uh, not gonna happen...
  
  --
  An enigma, wrapped in a riddle, shrouded in bacon and cheese
31. Re:I propose a test ... by Ichijo · 2014-05-22 09:06 · Score: 1
  
  I've been on many interstates. How is the I-294 different?
  
  --
  Any sufficiently unpopular but cohesive argument is indistinguishable from trolling.
32. Re:I propose a test ... by Ksevio · 2014-05-22 09:09 · Score: 1
  
  But computers can have multiple input sources and be monitoring all (potential) blind spots at once. Humans only have the forward facing eyes. Of course it remains to be seen if the software can handle the inputs to make good driving decisions, but that's not an impossible challenge.
33. Re:I propose a test ... by Ksevio · 2014-05-22 09:11 · Score: 1
  
  Even if you jammed the LIDAR, visual (or thermal) cameras could detect the cars moving. It's non-car objects like children or tree branches that are more difficult.
34. Re:I propose a test ... by CanHasDIY · 2014-05-22 09:12 · Score: 1
  
  I think he's basing that statement on the existence of "road nannies," i.e., the jackasses who will cut you off when you're in the passing lane and creep, thus creating a road hazard. Although, to be fair, I would hope the auto-cars would be programmed well enough to not change lanes if it detects another vehicle coming up fast.
  
  --
  An enigma, wrapped in a riddle, shrouded in bacon and cheese
35. Re:I propose a test ... by Joe_Dragon · 2014-05-22 09:12 · Score: 1
  
  with a line like No, people who drive above the speed limit put themselves at risk. you must not know that much about it.
36. Re:I propose a test ... by Anonymous Coward · 2014-05-22 09:16 · Score: 0
  
  I modded you down you ignorant piece of shit. Fuck you.
37. Re:I propose a test ... by ShanghaiBill · 2014-05-22 09:19 · Score: 3, Insightful
  
  That isn't any sort of a problem for LIDAR at sufficient resolution.
  Indeed. Google regularly subjects their cars to these types of scenarios, both in simulation and on test tracks. It is odd that when people try to come up with situations that SDCs "obviously" can't handle, they so often describe situations where SDCs have a clear advantage due to their much faster reaction time.
38. Re:I propose a test ... by Ichijo · 2014-05-22 09:20 · Score: 1
  
  What is different about the I-294 compared to other interstates?
  
  --
  Any sufficiently unpopular but cohesive argument is indistinguishable from trolling.
39. Re:I propose a test ... by Anonymous Coward · 2014-05-22 09:21 · Score: 0
  
  Tell that to my aunt, devotes a significant amount of her contextual awareness to what's on her facebook wall, even while driving.
40. Re:I propose a test ... by RockClimbingFool · 2014-05-22 09:21 · Score: 2
  
  Can you apply a different brake pressure to each one of your 4 wheels?
41. Re:I propose a test ... by FlyHelicopters · 2014-05-22 09:30 · Score: 1
  
  Humans can adapt to situations they have no prior experience with, usually after failing the first few times. Watch a 1 year old try and walk, for example.
  As for computers, they are stunningly good at doing things they have been programmed for.
  Airplanes can take off, fly, and land better than you or I can, far, far better... Airplanes have had autoland for almost 40 years now, people can't do that without being able to see something (even if it is using thermal or other enhanced technologies).
  In the big picture, computer driven cars will be much safer than human driven ones...
42. Re:I propose a test ... by epyT-R · 2014-05-22 09:31 · Score: 0
  
  This is not a troll statement. The premise was that the computer was safer than a driver who isn't aware of the terrain. My point was that a programmer designing for terrain he's unfamiliar with creates the same problem. The fact is I'd trust a human unfamiliar with mountain driving over a computer programmed by someone whoassumed a relatively flat city or suburban area, and who's not even in the car taking the same risks with us.
43. Re:I propose a test ... by ShanghaiBill · 2014-05-22 09:37 · Score: 1
  
  Depending on how well driverless cars handle that sort of thing (and how well they can trace any jamming back to the source), ...
  Self driving cars, or SDCs, use multiple sensors, including LIDAR, visible light cameras, GPS, rotary encoders and inertial sensors. The LIDAR is usually infrared, rather than the RF used in RADAR. Good luck jamming that. Since your jammer would have to be high powered direct line-of-sight, and SDCs record sensor data, tracing the source would be easy. Even if you shut down all the sensors, the car would apply the brakes, and pull off the road. It is conceivable that you could kill someone, but it would require quite a bit of technology, planning, and luck. It would be much easier to buy a gun and shoot them in the head as they drive past (in either a normal car or an SDC).
44. Re:I propose a test ... by zieroh · 2014-05-22 09:40 · Score: 1
  
  Screw passive cloaking, lets see how well that self driving sensor package works when getting hit full in the face with some broad spectrum jamming.
  
  Oh, it's a lot easier than that. I anticipate that we'll see exploits right off the bat that are based solely on specific behaviors next to / in front of / around an autonomous vehicle. It probably won't be hard to force one off the road just by aggressively encroaching into their lane and matching their rate of deceleration.
  Dangerous, sure. Irresponsible, absolutely. But inevitable just the same.
  
  --
  People who say "sheeple" have about as much sophistication as an AOL user, and in fact are probably actually AOL users.
45. Re:I propose a test ... by Jane+Q.+Public · 2014-05-22 09:42 · Score: 1
  
  That isn't any sort of a problem for LIDAR at sufficient resolution.
  Yes, it is. People keep insisting on thinking of just one small part of the issue as being the whole problem.
  
  What happens when it sees a vehicle coming the wrong way down the highway and there are cars in the adjacent lanes? How does it choose whether to smash into the car on the right, which has kids in it, or the one on the left, which is a big truck, or take the head-on?
  
  Just for example.
  
  This is probably what is going to happen, just like it happened to computer-controlled engines, antilock brakes, and other such things:
  
  One of these is going to kill somebody. There will be an uproar and lawsuits. The whole idea will be killed off for a while, then the tech will improve and it will come back. Except the same thing will happen again. And people will avoid them again. And they will come back.
  
  It's just that this is so vastly complex a problem, compared to leaning the fuel mixture and running your digital display, that it will go through many more of these cycles than things like antilock brakes did.
  
  And let's not forget: if the car "decides" to hit that car on the right, killing those kids, who is at fault? The programmer? The automobile manufacturer? Both?
  
  This is far from simply a "better hardware" problem.
46. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:02 · Score: 0
  
  Just like you fail at English.
47. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:05 · Score: 0
  
  If I had 4 pedals I could, smartass.
48. Re:I propose a test ... by Em+Adespoton · 2014-05-22 10:06 · Score: 1
  
  Even if you jammed the LIDAR, visual (or thermal) cameras could detect the cars moving. It's non-car objects like children or tree branches that are more difficult.
  I'd think thermal jamming would be the first thing someone would attempt -- it'll block both the thermal and visual cameras, as visual cameras always go a bit into the thermal spectrum and can be overloaded.
  LIDAR would actually be trickier, as you'd have to know exactly what range you were supposed to be jamming. You'd have to use a spread-spectrum scanner these days to even begin to figure that out.
  Personally, I think using a spread-spectrum scanner to plot the location of foreign objects would be an excellent first-level monitoring system on these cars. Fall back to LIDAR, and from there to visual/thermal. Or use all of the above for a composite image.
49. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:07 · Score: 0
  
  You're still fucking trolling
50. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:11 · Score: 0
  
  The difference is, they let dumb shits like Joe drive on their goddamned Interstates.
51. Re:I propose a test ... by Em+Adespoton · 2014-05-22 10:14 · Score: 1
  
  For that matter, computers have a reliable scheduler, so they can safely text and drive. I can let my car post status updates for me automatically without having to touch a screen....
52. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:15 · Score: 0
  
  Don't act like we don't read what you post in other articles. You advocate for people to dissent against laws they find unjust all the time. So yeah, he is objecting to that law. What was your point?
53. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:19 · Score: 0
  
  Broad spectrum jamming? Like shinning a bright light into the driver's eyes and honking?
54. Re:I propose a test ... by Em+Adespoton · 2014-05-22 10:21 · Score: 1
  
  When the normal speed of traffic is above the posted speed limit, self-driving cars will drive the speed limit as legally required but will cruise in the right lane as legally required when driving below the normal speed of traffic.
  So what happens when the right lane is an unmarked or badly marked exit lane?
  Personally, I think most of the collision avoidance stuff has been worked out now. It's the road navigation quirks that I'm more concerned with, as it seems it'd be pretty easy to force one of these off the road or into an exit lane. Just look at some of the mistakes professional GPS systems make; a car depending solely on these and immediate visual navigation are going to have a pretty hard time navigating complex shifting laneways and new routes.
55. Re:I propose a test ... by Anonymous Coward · 2014-05-22 10:22 · Score: 0
  
  There's no reason to special case this -- I'd expect it to *always* look in both directions when turning. Just like humans should.
56. Re:I propose a test ... by jez9999 · 2014-05-22 10:23 · Score: 1
  
  Don't get me wrong, I'm personally not interested in one of these self-driving contraptions
  Why not? They sound a fuckload better than the daily commute, if you still had the opportunity to drive yourself for pleasure when you felt like it.
  
  --
  == Jez ==
  Do you miss Firefox? Try Pale Moon.
57. Re:I propose a test ... by R3d+M3rcury · 2014-05-22 11:06 · Score: 1
  
  At night, during a blizzard, on a section of road with more holes than a chunk of Swiss cheese.
  Uphill! Both ways!
  I had to deal with that when I was a kid. But tell the driverless cars of today that, they'll just sit and stare at you.
58. Re:I propose a test ... by Copid · 2014-05-22 11:08 · Score: 3, Insightful
  
  I think that the driver of the vehicle going the wrong way down the highway would probably be at fault.
  
  It's easy to come up with vanishingly rare scenarios with no solution that a computer won't be able to solve (although how a human driver would do better in this situation is beyond me). Making policy based on bizarre edge cases is silly.
  
  --
  An interesting anagram of "BANACH TARSKI" is "BANACH TARSKI BANACH TARSKI"
59. Re:I propose a test ... by Obfuscant · 2014-05-22 11:30 · Score: 1
  
  Airplanes can take off, fly, and land better than you or I can, far, far better...
  
  Because airplanes have wings that provide lift and you and I don't.
  If you are saying that autopilots can fly aircraft better than you can, well, that may be true. I'd like to see one handle the Gimli Glider or Sully's Water Landing better than the pilots involved. Or the the Sioux City DC9.
  
  Airplanes have had autoland for almost 40 years now,
  Some airplanes have had that, and it requires special certification for the aircraft and crew to do it. A bit more training than the typical Department of Motor Vehicles road testing for a new driver's license. Many more airplanes do NOT have it, and many do not have any autopilot at all. Why? Because the cost of an autopilot is high based on liability issues when they get the pilots into trouble.
  
  people can't do that without being able to see something
  And autopilots in aircraft can't do it without being able to "see" something, either. They require a certified ILS or MLS approach (to "see" the vertical and lateral flight path), and a radar altimeter (to "see" the ground). And the Wikipedia entry for "Autoland" points out: "they are not generally smooth in their responses to varying wind shear or gusting wind conditions - i.e. not able to compensate in all dimensions rapidly enough - to safely permit their use." So a human pilot who could deal with a wind shear is better at it than the autoland autopilot.
  
  In the big picture, computer driven cars will be much safer than human driven ones...
  That may turn out to be true, someday, somewhere, in some instances. It is far from a fact, however. I view this claim using the hindsight of hearing the claims that we would all have helicopter-cars by 1980, and based on an understanding of how aviation autopilots have their limitations and have not become ubiquitous and mandatory.
60. Re:I propose a test ... by Obfuscant · 2014-05-22 11:33 · Score: 1
  
  as visual cameras always go a bit into the thermal spectrum and can be overloaded.
  Visual cameras have no thermal capability. Many of them are sensitive to near IR (if there are no explicit IR filters installed, which color cameras certainly have to keep the colors correct), but that's not what a thermal camera sees. Thermal sensors are in the far IR.
61. Re:I propose a test ... by pkinetics · 2014-05-22 11:38 · Score: 1
  
  Surprise surprise, auto manufacturers do test vehicles in harsh weather conditions.
  An autonomous driving vehicle is more likely to be subjected to extreme conditions, than the routine part like an ignition switch.
  I'd be more worried about downloading and installing JohnnyCab 2.1.3 update and whether or not they did proper regression testing. Testing? We didn't change any of that code, it'll be fine.
62. Re:I propose a test ... by Em+Adespoton · 2014-05-22 11:59 · Score: 1
  
  I fail to see how your response negates my statement. Even with IR filters, visible spectrum cameras can be overloaded with thermal energy. The main thing that gets overloaded is the focus logic that most visible spectrum cameras use, but the sensors themselves get too strong a signal for the filter to fully block, and that overloads the CCDs across the board, so that the signal to noise ratio is too near the floor.
  Dedicated thermal sensors as you say are far into the IR spectrum, and can be overloaded by the same IR burst, as IR tends to be noisy and travel across the waveform spectrum instead of being stable at a single frequency.
  If you've got a few cameras (thermal and visual) around, you can test this with an IR laser -- point it at each device while it's on, and check the result -- it will likely be a big white spot in the image.
  I caneven do this with my IR-based remote for my digital cameras -- it's not strong enough to really mess with the image, but it is strong enough to register as a white dot on most digital cameras when aimed directly at the lens.
63. Re:I propose a test ... by jopsen · 2014-05-22 12:01 · Score: 1
  
  I call it the aggressive, psychotic driver who makes random, unsafe lane changes, fails to signal, and swoops across several lanes of traffic while doing well over the speed limit.
  Lemme see your driverless car handle that, then we'll see.
  I personally look forward to driverless cars... no more looking before crossing the street..
  If you want to cross the street just jump out in front of a car...
  
  I saw one of the cars when I visited Mountain View a few weeks ago, and I was so tempted to jump out infront of it to see what happens :)
64. Re:I propose a test ... by jopsen · 2014-05-22 12:03 · Score: 1
  
  Yea, because the government totally won't have a backdoor already installed at the factory for dealing with... let's call them "undesirables."
  What makes you think they don't already have that... most cars can be drive to accidents electronically these days...
65. Re:I propose a test ... by amRadioHed · 2014-05-22 12:10 · Score: 2
  
  A computer controlled car can't do worse than a human in that sort of situation. Most drivers would probably just reflexively swerve somewhere before even seeing what is going on in adjacent lanes.
  
  --
  We hope your rules and wisdom choke you / Now we are one in everlasting peace
66. Re:I propose a test ... by amRadioHed · 2014-05-22 12:17 · Score: 1
  
  People have these same problems.
  
  --
  We hope your rules and wisdom choke you / Now we are one in everlasting peace
67. Re:I propose a test ... by dpidcoe · 2014-05-22 12:28 · Score: 1
  
  Exactly, just a lot more subtle and the responses will be easily replicable.
  
  If I want to see how shining my lights and honking at someone will affect their driving, I'd need to either do it to them personally or make a guess as to how they might act in response. If I want to see how it works with someones self driving car, all I need to do is obtain their model of car and an empty parking lot with which to test. Then if I find out through testing that the self driving cars reaction to having its LIDAR blinded with light outside of the visible spectrum is to hit the brakes and move to the right for example, and then find a spot where that exact response is exactly the wrong thing to do.
68. Re:I propose a test ... by dpidcoe · 2014-05-22 12:40 · Score: 1
  
  GPS
  Not gonna help you avoid road obstacles, or even do much more than really general navigation. Also jammable.
  
  rotary encoders
  Again, not going to help you much at avoiding things unless they're things that you saw before being jammed.
  
  inertial sensors.
  depends on how accurate they are, but that goes into the same category as rotary encoders in terms of avoiding things.
  
  The LIDAR is usually infrared, rather than the RF used in RADAR.
  Yeah, did the term "LIDAR" give it away?
  
  Since your jammer would have to be high powered direct line-of-sight, and SDCs record sensor data, tracing the source would be easy.
  It doesn't have to be very high powered at all, not does it need to be a brute force DOS attack. I bet there are plenty of ways of spoofing it to think there's something there that isn't, or otherwise give it bad data about an object. And tracing doesn't help you at all if it's a small object someone placed next to the road the night before your morning commute, and then collected again shortly after you crashed into a ditch.
  
  Even if you shut down all the sensors, the car would apply the brakes, and pull off the road.
  Which would really suck if there was actually no road there.
  
  It is conceivable that you could kill someone, but it would require quite a bit of technology, planning, and luck. It would be much easier to buy a gun and shoot them in the head as they drive past (in either a normal car or an SDC).
  A more apt analogy would be to say that it would be easier to saw most of the way through their tie rods or to poke holes in their brake cables. The point was to discuss this as a new angle of sabotage on top of the already crazy things that one can do with the electronics in modern cars, not to figure out the most effective way of killing someone.
69. Re:I propose a test ... by Obfuscant · 2014-05-22 12:45 · Score: 1
  
  Even with IR filters, visible spectrum cameras can be overloaded with thermal energy.
  Since they are not sensitive in the thermal (far IR) bands, they can't really be overloaded there. And the glass in the lenses pretty well stops the far IR, not to mention the visible cutoff filters. True thermal cameras don't have glass lenses.
  
  The main thing that gets overloaded is the focus logic that most visible spectrum cameras use,
  
  If the camera doesn't see the radiation, it won't impede the focus algorithms. But I would hope that the autonomous sensors are not auto-focus to begin with. They should be fixed focus.
  
  If you've got a few cameras (thermal and visual) around, you can test this with an IR laser -- point it at each device while it's on, and check the result -- it will likely be a big white spot in the image.
  I have both, and the near IR is stopped very well by the filters for the visible cameras, and the IR ones don't see near IR at all.
  
  I caneven do this with my IR-based remote for my digital cameras -
  Your IR remote is operating in the near IR, which is not where thermal cameras operate. Your near IR remote will have no effect on true thermal cameras, other than the heat generated by the LED itself.
70. Re:I propose a test ... by FlyHelicopters · 2014-05-22 15:02 · Score: 1
  
  If you are saying that autopilots can fly aircraft better than you can, well, that may be true. I'd like to see one handle the Gimli Glider or Sully's Water Landing better than the pilots involved. Or the the Sioux City DC9.
  You are quite right...
  However, if you add up all the accidents caused by human pilots and add up all the "saves" caused by human pilots, you're not ahead.
  If you went ahead and allowed everyone to die who otherwise would have been saved by a human pilot, yet saved everyone who was killed by a human pilot, you would be net ahead of the game.
  ---------
  Example... Airbags in cars have saved thousands of lives... they have also killed a few dozen people... do you remove them because of even one death, or do you accept that is the price of saving thousands of others?
  Autodriving cars will, sooner or later, kill someone. What about the thousands who didn't die because of them?
71. Re:I propose a test ... by ColdSam · 2014-05-22 16:39 · Score: 1
  
  You've made an excellent case for why you, personally, should not be designing a driverless car.
72. Re:I propose a test ... by Lennie · 2014-05-22 19:08 · Score: 1
  
  If someone runs that kind of software on the same computer driving the car, they are nuts.
  
  --
  New things are always on the horizon
73. Re:I propose a test ... by delt0r · 2014-05-22 21:22 · Score: 1
  
  It will handle it better than any slow to react wet computer with only two rather directional sensors.
  
  --
  If information wants to be free, why does my internet connection cost so much?
74. Re:I propose a test ... by Anonymous Coward · 2014-05-22 21:43 · Score: 0
  
  It would stop, though if you're going to try it, be sure it does have sufficient stopping distance to actually come to a halt before it reaches you. There is a video of one doing that when it was turning a corner at an intersection and someone walks out in front of it.
75. Re:I propose a test ... by swillden · 2014-05-23 00:12 · Score: 1
  
  Also, let's see how it deals with a bi-directional bike lane where one of the lanes is against traffic. These exist in Montreal, as well as other places I'm sure. Instead of putting 1 bike lane on each side, they put them both on one side of the street. When turning across the bike lane, you have to watch out for cyclists coming from both directions.
  Why in the world would you think that's a problem for a vehicle with LIDAR continuously scanning in all directions? Unlike a human who has to remember to look both directions, an automated car will always be looking in all directions.
  
  Also, let's test in snow. I'm tired of these things that only work in sunny California. It's bad enough that cell phones don't work with gloves on. I would hate it if my car failed to operate when there was a little bit of snow falling.
  Even if the computer couldn't handle driving in snow your car wouldn't "fail to operate", it would just fail to drive itself. You'd have to operate it, as you do now. One common concern about this situation is that your driving skills would atrophy and then be called upon in the worst conditions, but it's possible that you would be better at handling slick conditions if you didn't train yourself to assume the higher friction coefficients of clean, dry pavement.
  Actually, though, I don't see why a self-driving car would have a lot of trouble with snow and ice. It would have to be programmed for it, but the car is likely better able than you to determine how much traction it has.
  
  --
  Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
76. Re:I propose a test ... by AmiMoJo · 2014-05-23 00:33 · Score: 1
  
  Most of the accidents I see are due to people not being contextually aware. I'd have more faith in a computer that can look in every direction at once.
  
  --
  const int one = 65536; (Silvermoon, Texture.cs)
  SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
77. Re:I propose a test ... by canadiannomad · 2014-05-23 00:34 · Score: 1
  
  More likely scenario: It would cease acceleration and break at a speed and accuracy you could only dream of and pull into the pocket behind the truck or family and match the speed of the truck exactly enough to slowly pull away from its bumper. A maneuver that not within the abilities of your average human driver. (Though I'm sure there are some stunt drivers who could handle it.)
  
  --
  Hmm, the humour and sarcasm seem to have been be lost on you.
78. Re:I propose a test ... by i+kan+reed · 2014-05-23 00:54 · Score: 1
  
  My point is that the objection to the law is okay, but objection to machines that obey laws is stupid.
79. Re:I propose a test ... by dave420 · 2014-05-23 00:58 · Score: 1
  
  Your discrimination astounds me. Here you are claiming that Indians thousands of miles away can't code (which is patently false, given the scientific advances happening in India), and in a thread not too long ago you were railing against women for absolutely no reason apart from your own insecurities. Due to the ease at which you make generalisations of millions (or, in the case of women, billions) of people based on your own perverse and easily-disproved prejudices, you are either a wonderful troll, or a hateful pathetic human being. Pick one - your choice.
80. Re:I propose a test ... by GTRacer · 2014-05-23 01:00 · Score: 1
  
  It probably won't be hard to force one off the road just by aggressively encroaching into their lane and matching their rate of deceleration.
  Umm, how is running a SCD off the road that way any different than if I did that to you? How would you react differently?
  
  --
  Defending IP by destroying access to it? That makes sense, RIAA/MPAA. Go to the corner until you can play nice!
81. Re:I propose a test ... by Anonymous Coward · 2014-05-23 02:54 · Score: 0
  
  No, but humans tend to adapt to things faster.
  No they don't. Remember the plane that went down in the Hudson? During simulator runs, every other pilot that tried the scenario failed to return to the airport. Sully performed as well as any human could. However, during the same tests, the autopilot triaged the double-engine loss and reacted fast enough to return to the airport and land safely.
  A bad computer is worse than a bad human driver, but we've advanced far enough that a good computer is better than the best humans in the world.
82. Re:I propose a test ... by MrL0G1C · 2014-05-23 03:14 · Score: 1
  
  It may be able to look in every direction at once, the problem is that it barely has a clue about what it is looking at. My understanding is that to test the cars in mountain view, Google made a finely detailed 3d digital map of the whole area, why - because the algorithms are not very good at recognising things. So, will they make a similar digital map of every other road in the world they the cars might travel and how will cars deal with off-road.
  Motorways are pretty simple, unlike the streets of some old cities - this is where I would expect autonomous cars to have difficulties.
  
  --
  Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
83. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 04:32 · Score: 1
  
  Making policy based on bizarre edge cases is silly.
  Not when those polices have to be made. You missed the biggest point of my example.
  
  This isn't just a simple human scenario of jerking the wheel in reflex. The program has to be given instructions about what to do in such scenarios. And when it is: who is responsible if the outcome is less than ideal? The manufacturer? The programmer(s)? Who?
  
  This is not something that can be just left to chance. Somebody is liable. Policy must be made.
84. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 04:34 · Score: 1
  
  A computer controlled car can't do worse than a human in that sort of situation. Most drivers would probably just reflexively swerve somewhere before even seeing what is going on in adjacent lanes.
  You say that now, but what if it were your family members that got killed in that car on the right when it could have gone left and hit the truck instead?
  
  Yes, there will be lawsuits. Yes, somebody human will be found liable.
85. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 04:36 · Score: 1
  
  You've made an excellent case for why you, personally, should not be designing a driverless car.
  Glad I wasn't proposing to then. Instead, I plan to make the policies about who is liable when a driverless car makes a seriously fucked-up mistake. Which will happen.
86. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 04:38 · Score: 1
  
  More likely scenario: It would cease acceleration and break at a speed and accuracy you could only dream of and pull into the pocket behind the truck or family and match the speed of the truck exactly enough to slowly pull away from its bumper.
  It might be a more likely scenario, but it isn't the scenario I was talking about. The car is in traffic and it's typical big-city traffic; the car is blocked in on both sides. That was part of the whole point of the scenario.
87. Re:I propose a test ... by ColdSam · 2014-05-23 04:49 · Score: 1
  
  Because serious fucked-up mistakes never happen with human drivers. Are you by chance descended from a long history of luddites or are you just an opportunist looking to make a profit by causing countless injuries and death with your ridiculously ignorant FUD?
88. Re:I propose a test ... by Altus · 2014-05-23 05:36 · Score: 1
  
  well if you started doing that I might vary my rate of deceleration quickly in order to avoid you, or stomp on the gas and try to pull around you to the side. I might do any number of things that a computer might particularly because those things might be technically illegal. More importantly though, the car will do what it is programmed to do and nothing else. If you know its programming you can exploit it including getting it to do things that the designers did not actually intended but that are behaviors that emerge from unusual or unexpected inputs. Humans are less predictable. Not that you can't still run someone off the road.
  
  --
  "In America, first you get the sugar, then you get the power, then you get the women..." -H. Simpson
89. Re:I propose a test ... by Copid · 2014-05-23 05:39 · Score: 1
  
  That's the thing, though. It's not "left to chance" in the sense you're talking about. In cases like that, the car likely does have very sensible, well-defined instructions that are dead nuts correct for that family of problems. It's just that those instructions are optimal for that family of cases rather than that specific case. If you're boxed in and you absolutely have to swerve into somebody, the best thing to do is probably to swerve away from oncoming traffic (to the right in the US), all else held equal.
  
  You adopt the strategy that will crack the fewest eggs over the long run, even if it doesn't account for things like, "How many children are in the car on the right?" or, "Is the childless guy on the left driving to see his mother on her deathbed for he last time?" or, "Is one of the children in the car on the left going to grow up to be the next Hitler?" Those factors may make one particular situation a little bit more interesting, but not interesting enough to hold the engineering firm responsible. Back to your example, unless the car decided to go on a wild and unnecessary killing spree because of the novel situation, the person who would be held responsible is the truck driver going the wrong way down the highway.
  
  A more general question that ends up being asked in the real world a little more often: If a car company can, say, move the gas tank in a way that reduces deadly fires but slightly increases the probability of a passenger's beheading, and that move will behead 1 person per year but save 10 per year from burning to death, should the family of the 1 person be able to sue them for making that change? Or do we acknowledge that strapping ourselves into boxes on wheels that go zipping down obstructed highways carries some inherent risk and that no engineering solution can optimally handle every situation?
  
  --
  An interesting anagram of "BANACH TARSKI" is "BANACH TARSKI BANACH TARSKI"
90. Re:I propose a test ... by Altus · 2014-05-23 05:45 · Score: 1
  
  And those people are held responsible for their fucked up mistakes. Who is responsible when the car makes a mistake. I sure as hell don't want to be financially responsible for some companies fuck up.
  This isn't about FUD its a legitimate question of insurance policy and law that will have to be addressed if autonomous vehicles are going to become commonplace. They are not unsolvable problems but neither can the be ignored by sticking ones head in the sand the way you clearly would like to do just because "omg self driving cars... COOL!"
  
  --
  "In America, first you get the sugar, then you get the power, then you get the women..." -H. Simpson
91. Re:I propose a test ... by ColdSam · 2014-05-23 06:19 · Score: 1
  
  And those people are held responsible for their fucked up mistakes.
  Really, it's that simple? When there are multiple cars involved, a myriad of non-optimal decisions and a fatality results it's trivial just to put the blame on one particular person?
  
  Who is responsible when the car makes a mistake. I sure as hell don't want to be financially responsible for some companies fuck up.
  Then you must not be a driver, because you put your life in their hands every time you get behind the wheel. You can not just will your brakes not to fail or your engine not to turn off because of a faulty ignition.
  
  This isn't about FUD its a legitimate question of insurance policy and law that will have to be addressed if autonomous vehicles are going to become commonplace.
  
  Nobody says there aren't serious questions of insurance and law that need to be resolved (and which will evolve over time). That's a big part of what this article is about: rules and regulations, liability, etc. Where the FUD comes about is these absurd scenarios from ignorant commenters like this: OMG a car is coming right at you, do you kill the children or sacrifice yourself!!!! Ignoring that these cars, by the time they are widely deployed, will be far safer than human driven cars, not only in the vast majority of ordinary boring cases, but even in most of these extreme and incredibly rare scenarios that JQP imagines. That's FUD.
  
  They are not unsolvable problems but neither can the be ignored by sticking ones head in the sand the way you clearly would like to do just because "omg self driving cars... COOL!"
  The issue is that most of the "problems" that luddites like JQP throw out are already solved (better than human drivers) and they are just too dumb, ignorant or lazy to consider them.
  I don't know, maybe there is a middle ground between sticking your head in the sand and screaming "OMG. Our cars are trying to kill us and nobody is stopping them. Won't somebody think of the children!" Just saying.
92. Re:I propose a test ... by Obfuscant · 2014-05-23 07:50 · Score: 1
  
  Autodriving cars will, sooner or later, kill someone. What about the thousands who didn't die because of them?
  Your argument sounds very much like "what about the thousands of terrorist bombs that didn't make their way onto airplanes because of TSA?"
93. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 08:35 · Score: 1
  
  Really, it's that simple?
  Nobody said it was simple. But you're being dense.
  
  Where the FUD comes about is these absurd scenarios from ignorant commenters like this
  If you want to have an "ignorance" contest, you probably picked the wrong person to insult.
  
  The issue is that most of the "problems" that luddites like JQP throw out are already solved (better than human drivers) and they are just too dumb, ignorant or lazy to consider them.
  There you go. You just lost the ignorance contest. I am about the farthest thing from a "Luddite" you are likely to find. I want to solve these real problems so that such technology can exist and be used in the real world.
  
  The scenario was an EXAMPLE of the kind of situations that can and do arise. It doesn't matter if they're "unlikely". Getting hit by lightning is unlikely, too, but it happens an average of 51 times every year in the U.S. The scenario was deliberately made to be an edge-case in order to illustrate a point. Several points, actually. But one of them is that when you're a programmer, like I am, you can't just leave such edge-cases alone and say "That will never happen." Because they do.
  
  It is people who are ignorant of how it actually works who think "it will never happen" is a real-world scenario.
94. Re:I propose a test ... by FlyHelicopters · 2014-05-23 09:10 · Score: 1
  
  Actually, no... I would reverse that...
  For every terrorist bomb found, millions of people were needlessly hassled every day by the TSA.
  Imagine if airplane hijackings were not reported on the news? No terrorist would bother, what would be the point.
95. Re:I propose a test ... by ColdSam · 2014-05-23 09:24 · Score: 1
  
  Nobody said it was simple. But you're being dense.
  Yeah, you kind of did. You are using the example of this incredibly rare and complicated self-driving car scenario as an example of what will go wrong, while just glossing over the also incredibly rare and complicated current examples. In your scenario how is a human driver better, how is the law simpler. It's not.
  
  If you want to have an "ignorance" contest, you probably picked the wrong person to insult.
  If you're not ignorant then you're just plain dumb or disingenuous. Take your pick.
  
  There you go. You just lost the ignorance contest. I am about the farthest thing from a "Luddite" you are likely to find. I want to solve these real problems so that such technology can exist and be used in the real world.
  That is not what you're doing at all. You are muddling and obfuscating the issue and trying to make it sound more complex, scary and unsolvable than it is. You're making unfair and unrealistic comparisons and sowing FUD, e.g. when you throw out the ridiculous "what if your kids were in the car?!" That shows you're not serious about examining, let alone solving any problems.
  
  The scenario was an EXAMPLE of the kind of situations that can and do arise. It doesn't matter if they're "unlikely".
  
  Sorry, but your scenario was ridiculously stupid. Of course it matters how unlikely the situation is. For every situation like you proposed there will be 10,000 times as many cases where another car is stopped in the road and it's at night or there's fog or the driver is inattentive and the self-driving car will save lives. There will 1,000 times as many cases where the head on collision is easily avoided because the car is not completely boxed in. Even if you do not put any extra clever coding the self-driving car will probably do better than a human driver. Once again, your example is stupid to be overly worried about.
  But even if 1 in 1,000,000 driverless cars just spontaneously combusted and killed every passenger it would still be worth having them on the road if they saved 10,000 lives a year.
  
  Getting hit by lightning is unlikely, too, but it happens an average of 51 times every year in the U.S.
  So maybe you should start preaching about that instead.
  
  The scenario was deliberately made to be an edge-case in order to illustrate a point. Several points, actually. But one of them is that when you're a programmer, like I am, you can't just leave such edge-cases alone and say "That will never happen." Because they do. It is people who are ignorant of how it actually works who think "it will never happen" is a real-world scenario.
  However, if you're a programmer with a job you know that for almost every real world system you will not be able cover every single case and if you try to write special code for every single imaginable possibility that your code will unmaintainable and untestable. If you write solid simple code that covers the vast majority of cases that is often good enough to get the job done. In the scenario you outlined you don't have to write a headOnCollisionWhileBoxedInBetweenATruckAndCarWithTwoKids() routine. Most likely the normal collision avoidance algorithms would be better than the current alternatives: ideally it would detect the head on car far enough away that it can slow down to get behind one of the vehicles or even slow down to take the hit. You certainly could work on routines to rank objects by the expected cost of running into them, but doing that well is extremely low down the list. Not to mention that even 1 of the other 3 vehicles was also self-driving then this 1 in 1,000,000 accident would never happen in the first place.
96. Re:I propose a test ... by ColdSam · 2014-05-23 09:27 · Score: 1
  
  It's nothing like that actually. One of those things is easily measurable, the other is not.
97. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 09:34 · Score: 1
  
  Yeah, you kind of did. You are using the example of this incredibly rare and complicated self-driving car scenario as an example of what will go wrong,
  Yes. One in a million, maybe? Well, the odds of getting hit by lightning are less than 1 in 6,000,000. Yet it happens, surprisingly often.
  
  ... while just glossing over the also incredibly rare and complicated current examples. In your scenario how is a human driver better, how is the law simpler. It's not.
  I didn't "gloss over" anything. I repeat: my example was deliberately chosen to be one of those edge-cases, in order to illustrate a point. But THE POINT has passed entirely over your head, and instead you are concentrating on the example I used to illustrate it. Just to demonstrate:
  
  For every situation like you proposed there will be 10,000 times as many cases where another car is stopped in the road and it's at night or there's fog or the driver is inattentive and the self-driving car will save lives.
  So what? That has nothing to do with THE POINT I was making. Which you still don't get.
  
  The real-world scenarios in which car computers were accused of causing accidents, and in which anti-lock brakes were accused of causing accidents WERE real, DID happen, DID result in lawsuits, and DID affect the markets for these devices for a long time. And the entire time, they were probably saving thousands of lives. Again: so what? That doesn't mean those other things didn't happen.
  
  At NO time in this entire conversation did I say that was a valid reason for this technology to not be used, or not be developed. On the contrary, I think it's a great thing, and will do the world a lot of good.
  
  But denying that there will be real problems along the way, and failing to use clear and recent history of an EXAMPLE of the kinds of problems that will almost certainly occur, is stupid.
98. Re:I propose a test ... by ColdSam · 2014-05-23 10:35 · Score: 1
  
  Yes. One in a million, maybe? Well, the odds of getting hit by lightning are less than 1 in 6,000,000. Yet it happens, surprisingly often.
  One in a million what? Are you saying that 300 times a year there is a case of head on collision where a driver is trapped between two other vehicles and must choose between your 3 options (I'll even ignore the extra detail that one is a truck and one has children). That's absurd. The likelihood of one of the few self-driving cars running into this exact situation is so close to zero that it's not worth spending a lot of time on. Sure you can run it through the simulator and put it in your regression tests, but it's really hard to imagine that it's going to require a lot of extra work to make workable.
  
  I didn't "gloss over" anything. I repeat: my example was deliberately chosen to be one of those edge-cases, in order to illustrate a point. But THE POINT has passed entirely over your head, and instead you are concentrating on the example I used to illustrate it. Just to demonstrate:
  Your example did not make any point, unless you were trying to make the point that even in rare cases self-driving cars are likely to be better than human equivalents. There are cases right now that driverless cars will make terrible, life-threatening decisions, many of which human drivers can perform better at). Some of these scenarios are going to be extremely difficult or next to impossible to solve in the near future (at a reasonable cost). Your scenario was not even close to being one of these.
  
  So what? That has nothing to do with THE POINT I was making. Which you still don't get.
  Perhaps I misjudged your intent, but only because THE POINT you are making is kind of facile and obvious. No reasonable person expects self-driving cars to be on the road in the next couple of years with no faults and no liability issues. Your tone suggests that these problems are nearly intractable and yet your scenario and concerns border on the trivial, IMO. Cars already have many "self-driving" features and they are possibly getting rolled out slower than they should because of these type of concerns, but they are still coming. Legislators and car manufacturers should be aware of history, but it shouldn't paralyze them and your hysterical tone, IMO, doesn't help.
  
  At NO time in this entire conversation did I say that was a valid reason for this technology to not be used, or not be developed. On the contrary, I think it's a great thing, and will do the world a lot of good.
  Reread what you've said, it has all been negative. You do not say that some current system might temporarily have issues with their LIDAR adequately handling super aggressive drivers. You just say it's a problem, full stop.
  
  But denying that there will be real problems along the way, and failing to use clear and recent history of an EXAMPLE of the kinds of problems that will almost certainly occur, is stupid.
  Good thing nobody is saying that then. This is not a productive conversation, I'm done.
  If you truly do think that self-driving cars will save lives then I recommend you change your tone to reflect that, rather than appearing to put your support behind idiots and fear mongers who say things like "What if your children were in that driverless car that ran amok?!" Your kind of talk is what makes otherwise sensible people do stupid things like drive across the country because planes are too dangerous.
99. Re:I propose a test ... by Jane+Q.+Public · 2014-05-23 12:54 · Score: 1
  
  Good thing nobody is saying that then. This is not a productive conversation, I'm done.
  You continue to argue with me over things I wasn't saying or even implying. So I'm glad you're done. I agree with you 100% that it has been an unproductive exchange.
  
  Reread what you've said, it has all been negative.
  You still obviously didn't get my original point. So I'm happy you're going away. You're wasting everybody's time.
100. Re:I propose a test ... by jopsen · 2014-05-23 19:56 · Score: 1
  
  It would stop, though if you're going to try it, be sure it does have sufficient stopping distance to actually come to a halt before it reaches you.
  Well, that's just boring... He he :)
  
  I see a future where teenagers jump infront of cars to see who dares wait the longest :)
101. Re:I propose a test ... by strikethree · 2014-05-24 09:24 · Score: 1
  
  What happens when it sees a vehicle coming the wrong way down the highway and there are cars in the adjacent lanes? How does it choose whether to smash into the car on the right, which has kids in it, or the one on the left, which is a big truck, or take the head-on?
  
  Most likely do the exact same thing I would do: Reduce speed and stay in lane.
  Why should an emergency that affects me also affect the people next to me? It does not matter if there are innocents or gang bangers to the left and right of me. I reduce speed and stay in my lane.
  Other than the moral implications of purposefully smashing into the cars next to me, there is another consideration: Just because the car is coming straight at you now, that does not mean it will continue to do so. What if the oncoming vehicle breaks right to squeeze in between you and the person next to you? Breaks left? You do not know what will happen.
  To be more frank, for the past 7 years, I have been dealing with EXACTLY this situation (long story about convoys and two lane roads) on a fairly regular basis. Everyone, so far, that has tried to avoid the head on collision has died a horrible death. Only one person, so far, who has just went ahead and accepted the inevitable has died. Six others have survived; although I am unsure of any permanent injuries. :/
  In conclusion, trying to avoid a head on collision is stupid. You will be MUCH worse off if your inertia is rotating along an axis that is not aligned with your direction of travel. Just reduce speed. This is an easy one for an autonomous car; it is not afraid of death. You are. You will do something stupid like trying to avoid the collision.
  
  --
  "Someone needs to talk to the tree of liberty about its ghoulish drinking problem." by ohnocitizen
Advanced Cruise Control by Anonymous Coward · 2014-05-22 07:58 · Score: 0

Until I can buckle up, hit "go" and take a nap from my driveway to my work parking spot...all this technology is just advanced cruise control. These new technologies are fantastic though - it seems like with some I can merge into rush hour and just let autopilot take over. It seems like it will be a few decades before it's truely automated though...to let the politics and insurance industry figure themselves out.
1. Re:Advanced Cruise Control by Russ1642 · 2014-05-22 08:13 · Score: 3, Insightful
  
  Technology needs to develop in stages. It doesn't go from concept to full-blown earth-shattering product like it does in bad science fiction movies.
2. Re:Advanced Cruise Control by jellomizer · 2014-05-22 08:18 · Score: 1
  
  Having taken long trips on the highway. This partial automation of keeping the speed limit, stay in lane, don't run into the other cars. Would be a welcomed feature. After 3 or 4 hours. Your eyes get strained from paying attention and so alert. Having this feature where you can sit back for and let your guard down for a bit would make my life so much better.
  
  --
  If something is so important that you feel the need to post it on the internet... It probably isn't that important.
3. Re:Advanced Cruise Control by Anonymous Coward · 2014-05-22 09:22 · Score: 0
  
  Modern cars have ~10,000,000 lines of code already, for traction control amongst many other things.
Translation by Anonymous Coward · 2014-05-22 08:03 · Score: 0

"permits" and "special licenses" == "fees"
1. Re:Translation by Joe_Dragon · 2014-05-22 08:11 · Score: 1
  
  well do you want there to be a victims fund or do you want to be the guy who get's hit buy a auto car and then have billing piling up when the lawyers and courts are talking there time fighting over who is at fault and who will pay up.
2. Re:Translation by Todd+Palin · 2014-05-22 08:18 · Score: 1
  
  I want to know who goes to jail if someone is killed under circumstances where the car makes a really stupid error that meets the standard for vehicular manslaughter?
3. Re:Translation by Talderas · 2014-05-22 08:24 · Score: 1
  
  Hence the $5m bond. The license for the "driver" is a bit weird.
  
  --
  "Lack of speed can be overcome. In the worst case by patience." --Znork
4. Re:Translation by Joe_Dragon · 2014-05-22 08:29 · Score: 1
  
  Movie idea there.
  Some who owns auto drive car ends up in having the car that due some fault ends up killing someone and due to all kinds of fine print, NDA's, Eula's is found at fault / the courts are working off older laws from before auto drive cars. The owner does some prison time. After getting out finds the only jobs they can get are mcjobs after doing that for some time Says the prison life was better so he go plan to down to Google HQ to do some big vandalism to both get revenge and to get back to the prison life.
5. Re:Translation by CrimsonAvenger · 2014-05-22 09:12 · Score: 1
  
  And this movie would be entertaining to whom?
  I mean, maybe as a comedy, possibly....
  
  --
  
  "I do not agree with what you say, but I will defend to the death your right to say it"
6. Re:Translation by Em+Adespoton · 2014-05-22 10:30 · Score: 1
  
  You were part way there... driver was a traffic cop, and the car death was a setup framed by evil villain X who traffic cop had given one too many tickets. When driver gets out, they find they're the only person who still knows how to drive, and the autonomous cars no longer listen to humans' instructions. The local government asks said ex-con to override a car and drive into navigation HQ to take out the mystery man/AI controlling the car network -- of course it's actually being managed by evil villain X.
  Let's call the movie "Demolition Derby"
Why so much insurance? by Anonymous Coward · 2014-05-22 08:08 · Score: 0

I'm pretty sure that normal cars aren't required to have a $5 million insurance policy in any state. So why require the automated cars to have a $5 million policy?
1. Re:Why so much insurance? by Russ1642 · 2014-05-22 08:11 · Score: 1
  
  When people find out they were hit by an autonomous with a driver who is only monitoring the system they're going to sue the crap out of the owner, driver, AND the state. And they damn well should.
2. Re:Why so much insurance? by jcochran · 2014-05-22 08:14 · Score: 1
  
  I'm pretty sure that the $5 million policy is for accidents caused by the vehicle while testing. AKA.... Unproven technology. Once all the tests have been passed, the insurance requirements for the general public would be more in line with the the insurance requirements for non-autonomous vehicles. And I suspect that since the autonomous vehicles would have a lower accident rate, the insurance premiums would be lower as well.
3. Re:Why so much insurance? by Tanktalus · 2014-05-22 08:17 · Score: 1
  
  Given that this is SCARY and NEW TECHNOLOGY, I can see an abundance of caution here. Also, it's the manufacturer that has to have the insurance, which seems to me to be rather cheap, especially since many players, especially Google, could self-insure something like this and wouldn't really notice any pinch. To be honest, this seemed to me to be somewhat low if their primary purpose is to ease peoples' minds about the new technology.
  Remember that the state is going down uncharted waters here, regulating these things prior to actual use as opposed to the catch-up-with-existing-practices we did the first time round with these horseless carriage things. So they're taking things easy. It's probably the best way to make everyone comfortable with the process, other than perhaps the manufacturers.
4. Re:Why so much insurance? by LifesABeach · 2014-05-22 08:19 · Score: 1
  
  It does make sense; your going to need to a lot of money for counseling for the Engineers when they are told their design is a heap of twisted metal at the bottom of some cliff.
5. Re:Why so much insurance? by gurps_npc · 2014-05-22 08:25 · Score: 4, Insightful
  
  Why exactly should they? Keep in mind that when they are hit by a drunk driver they have to settle for a small insurance payout.
  Your basic assumption are flawed. Specifically you appear to believe (with zero reason for these beliefs) that:
  An autonomous car will be less safe than the average driver - including drunks, teenagers, old people and parents screaming at their kids to stop fighting.
  That the owner should somehow be responsible in ANY way, rather than the programming company.
  That the 'driver' - who in this case is not driving but instead monitoring to make sure the system is working - is in anyway responsible.
  That the state would let autonomous cars drive if they were not proven better drivers than teenagers and senior citizens.
  I predict that the states will have reasonable requirements for the autonomous car to prove that they are SIGNIFICANTLY better drivers than your average teenager - as in perfect driving with zero mistakes on driving tests that are far more difficult than what people have to pass.
  I then predict that they will force the programming company to accept all legal responsibility for driving when the system is working - and the owner/monitor will only be required to verify that the vehicle has no warning lights activated.
  I further predict that car accidents will drop to a tenth of what it is currently - and that video recordings taken by the autonomous cars will prove that except in the rare circumstances they are in an accident is because some human violated driving laws, not that the car made bad decisions
  Driving is something that an artificial intelligence should be able to do far better than a human. Their reflexes are better, they don't get angry, they don't get impatient, they don't get drunk, they don't get worse as they age and they start out experienced rather than as a beginner.
  Anyone that can't understand sounds like a Luddite to me.
  
  --
  excitingthingstodo.blogspot.com
6. Re:Why so much insurance? by epyT-R · 2014-05-22 08:28 · Score: 1
  
  Because computers have a contextual awareness roughly equivalent to a cat? ..at best
7. Re:Why so much insurance? by Russ1642 · 2014-05-22 08:29 · Score: 1
  
  When I was working at a major oil processing facility I had to have $2 million insurance just to drive on site and park in the lot. A normal policy is $1 million. Asking for $5 million dollars in liability insurance isn't outrageous.
8. Re:Why so much insurance? by gurps_npc · 2014-05-22 08:56 · Score: 1
  
  Your company paid the $5 million as a working expense - so it reduced their tax burden. But eventually this should end up being in all cars - including those owned by the poor.
  I could see the first set of cars - being owned by the wealthy and by corporations - having $5 million insurance.
  But we are talking about the final situation when the cars are sold to customers, vs being 'tested'. That requires the technology be proven to be less likely to cause accidents than human drivers, and I expect the insurance to be switched. That is, autonomous cars would have normal insurance ($1 million or whatever the state deems normal) and and human driven cars would be required to have excessive insurance ($5 million for example) and also pay a lot more for the same insurance.
  If the car isn't safer than your average human driver, it won't be approved for general use. If it is safer than they should pay less for their insurance.
  
  --
  excitingthingstodo.blogspot.com
9. Re:Why so much insurance? by onkelonkel · 2014-05-22 10:30 · Score: 1
  
  Just to be clear, we are talking about a $5 million limit to a liability insurance policy. It doesn't cost $5 million. I pay about $250 a year to raise my minimum $200k liability to $3 mil. Apparently I'm a bit above the local average, most people here have about $2mil coverage. If I had more personal assets, and I lived in a more litigious place I would probably go for the max $5 mil .
  
  --
  None of them can see the clouds; The polished wings don't care.
10. Re:Why so much insurance? by HiThere · 2014-05-22 12:32 · Score: 1
  
  And if the cars prove as much safer than human drivers as has been predicted, the cost of a $5 million insurance policy on them (while driven by the robot) might well be cheaper than $1 million while driven by a human.
  
  --
  
  I think we've pushed this "anyone can grow up to be president" thing too far.
11. Re:Why so much insurance? by Anonymous Coward · 2014-05-22 22:02 · Score: 0
  
  I think you are vastly underestimating the awareness of cats here.
12. Re:Why so much insurance? by Anonymous Coward · 2014-05-23 11:51 · Score: 0
  
  And I predict that loonies of twin towers fame will appropriate said driverless cars and load them with outrageous contents....
  Sorry but a certain fraction of 'humanity' will exploit technology just as they did aircraft. Then how will autonomous cars fare?
  'Humanity' is not yet ready to let these things loose. Insurance is trivial in comparison.
So what's the California RMV? by rsborg · 2014-05-22 08:20 · Score: 1

I assume it's the Rancho Mission Viejo?

--
Make sure everyone's vote counts: Verified Voting
1. Re:So what's the California RMV? by aicrules · 2014-05-22 09:00 · Score: 1
  
  California Registry of Motor Vehicles
2. Re:So what's the California RMV? by Anonymous Coward · 2014-05-22 12:12 · Score: 0
  
  California Registry of Motor Vehicles
  Nice try, but there is no such institution. Automobile licensing in the state of California is
  governed by the Department of Motor Vehicles. As this thread's initiator was gently pointing
  out, the summary has got it wrong.
A Special Drivers License? by LifesABeach · 2014-05-22 08:22 · Score: 1

I hope it's not the same special training that Law Enforcement gets for driving at 70+mph without a seat belt on. I think the lesson on sudden impacts loses a lot of students.
Autopilot with alarms by Anonymous Coward · 2014-05-22 08:24 · Score: 0

If I could kick it back and read during rush hour, that would be awesome. All it would need is some kind of "DANGER!" alarm to quickly get my attention if something bad were about to happen.
Guess engineering better hurry by russotto · 2014-05-22 08:30 · Score: 1

...because it looks like regulation is working overtime to stifle development of autonomous cars before they become practical.
Why not? by Anonymous Coward · 2014-05-22 08:57 · Score: 2, Insightful

Don't get me wrong, I'm personally not interested in one of these self-driving contraptions, ...
I would LOVE one of these things. I hate driving and I can't afford to hire a driver. And after an 8 hour road trip yesterday, I would have LOVED to have an auto driven car. Hang back read or something. Because towards the end of the trip, I was having a real hard time concentrating.
But if I have to pay attention even if under automatic control, then I don't see the point. If I have to pay attention, then I might as well do the driving myself.
1. Re:Why not? by FlyHelicopters · 2014-05-22 09:27 · Score: 3, Insightful
  
  But if I have to pay attention even if under automatic control, then I don't see the point. If I have to pay attention, then I might as well do the driving myself.
  Understandable reaction, but you're wrong.
  Autopilots in airplanes do not remove the pilot's requirement to pay attention to what is going on. In fact, by NOT having to physically fly the plane, the pilot has a better idea of what is going on around him/her.
  I can tell you from much experience that autopilots are wonderful things, you'll see more and be aware of so much more once the car does the driving.
2. Re:Why not? by zieroh · 2014-05-22 09:38 · Score: 2, Insightful
  
  Understandable reaction, but you're wrong.
  The AC was presenting his personal opinion -- that if he has to pay attention, he's not interested. Technically speaking, one cannot be "wrong" about whether he is interested in it or not. He's either interested, or he's not, and there's only one person on the entire planet that actually has a say in that.
  
  --
  People who say "sheeple" have about as much sophistication as an AOL user, and in fact are probably actually AOL users.
3. Re:Why not? by Anonymous Coward · 2014-05-22 09:48 · Score: 0
  
  If I have to pay attention, then I might as well do the driving myself.
  That's where they were wrong.
4. Re:Why not? by Obfuscant · 2014-05-22 11:11 · Score: 3, Informative
  
  Autopilots in airplanes do not remove the pilot's requirement to pay attention to what is going on.
  
  Autonomous vehicles are being promoted for exactly their ability to allow the driver to do non-driving functions (like read, eat, nap, or other things). That's what they mean when they talk about increasing productivity of those who commute to work using one.
  
  I can tell you from much experience that autopilots are wonderful things,
  Yes, they are. But they are not intended to allow flight in close formation (like a string of autos on the freeway would be), or in close proximity to the ground*. These autopilots are intended for an environment where the closest thing to you is more than 500 feet away at least. And they will quite happily fly you into the ground when they fail. Or fly you to the point you stall and then fall to the ground.
  Here's just one example of how the autopilot can fail, even though the NTSB would call it pilot error (just like everything else, almost.) The Garmin G1000 with (mumble) 700 autopilot has a VS command. That's "vertical speed". You can tell the autopilot to climb at a set rate, say 500 fpm. If you forget to add power you may not be able to achieve a 500 fpm climb (or you may initially make it, but as you climb the performance decreases and you can't keep it) -- but the autopilot will keep trying. It will try to increase your angle of attack to get more lift so you can climb at the rate you've requested. It will keep trying so hard that it may cause your airspeed to drop below stall speed.
  Hello, Pilot, you are now in a full-on stall, probably about to enter a spin, maybe in IMC, and your autopilot has adjusted your elevator trim to full-up trying to do what you told it to do. It's your aircraft. HAND.
  It has, in it's data, the stall speeds for the aircraft it is in because it will display the critical speeds as flags on the airspeed indicator, so it could easily report the problem to the pilot. "Bong -- minimum airspeed reached, climb aborted." It does not. There is a recent article in Aviation Safety, I think it was, about a crash of a military version of a King Air in the mideast that did exactly that. The pilot was IFR and distracted and the aircraft stalled and then spun in.
  There is a very good reason that there are half a dozen (8 for the G1000, as I recall) or more ways of disabling the autopilot in an aircraft. They fail often enough, and in serious enough ways ("hey, let's run your elevator trim FULL UP for no reason at all, bud", e.g.) that it is important to be able to kill George immediately. And have multiple ways to kill him in case the first three didn't kill him well enough.
  You should probably not use aircraft autopilots as examples of robo-perfection.
  * Yes, there are cat III autopilots that will fly the aircraft onto the runway, or "auto-land". It takes specially certified equipment AND CREWS to do that. You aren't going to find many Joe Sixpacks out on a drive that could meet equivalent quals. And that G1000 I use as an example? The aircraft manual prohibits use of the autopilot below 800 AGL, as I recall.
5. Re:Why not? by FlyHelicopters · 2014-05-22 14:38 · Score: 1
  
  You make a good point, fair enough, if the OP doesn't want one because of his personal views, then he is not "wrong" in his opinion.
  I of course was referring to the point that just because you have to pay attention doesn't make an autopilot useless, it actually enhances safety.
6. Re:Why not? by FlyHelicopters · 2014-05-22 14:59 · Score: 1
  
  Autonomous vehicles are being promoted for exactly their ability to allow the driver to do non-driving functions (like read, eat, nap, or other things).
  Ahh, fair enough... I think we're further away from *that* than we are autodriving cars where you still have to pay attention.
  The car you describe, I could put my kids into and it would take them to school, without my even being there. I think we're more than a few years away from THAT. :)
  
  But they are not intended to allow flight in close formation
  Actually, you might be surprised, there are such autopilots... First, the autopilot on most modern airliners can do a Cat III autoland in zero/zero conditions. You don't touch the controls until you're doing a go around. The autopilot controls the airplane all the way to landing and roll out, usually to below 80 kts, then you can take over.
  The Navy's airplanes have autoland to the carriers, which is even harder than to a fixed airport (since the runway is moving in three dimensions).
  And finally, some Air Force autopilots have master/slave modes, you can link them to fly in formation long distances together. The master has the plan and the slaves simply hold position on the master.
  
  And they will quite happily fly you into the ground when they fail. Or fly you to the point you stall and then fall to the ground.
  Some will, others will not. The example in the small airplane with the Garmin G1000 will because it lacks FADEC and autothrottles. An Airbus A320 won't let you do that, it will prevent you from pitching over 33 degrees up or down and rolling more than 67 degrees left or right. It will also prevent a stall by overriding your throttle setting and applying more power, then lowering the nose if you continue to try to pull back the stick and full throttle has been reached.
  You can manually override that in an emergency (that was added after the Paris Airshow crash many years ago of the A300, that was pretty stupid on the pilot's part) so that you can roll it upside down if you really need to, but normally the plane won't let you do that.
  As for flying into the ground, modern airliners won't let you do that anymore either, besides having GPWS (ground proxomity warning system) among other things, the new ones will override the controls and avoid a collision with the ground unless in landing mode and facing an airport. Check out the G550 and G650 some time, they have amazing computer systems in the cockpit, including FLIR and CCIP, along with verbal callouts.
  http://youtu.be/lJIvsI9AtIs
  This one is a Gulfstream G450 landing at Aspen
  http://youtu.be/DR9lyAM2YNE
  This one shows taxing on the ground and other shots of the enhanced vision system.
  If you watch the first one, you'll see a small circle with a line coming out of the sides and top, that is the CCIP (continuously computed impact point), that shows where the airplane will go if you do nothing else, so long as it is on the end of the runway, that is where you'll end up. If it is showing lower or off to the side or in the trees, you better do something else.
  ----------------
  Side note: Yes, I'm aware that at the end, you pointed out that the autopilot in a C172 and a G1000 is not a great example because that system is stupid. Yes it is, because of the cost of development in small airplane aviation is just sad and behind the times, that is what you get. Because of the large number of cars built, billions can be spend on development, I would expect anything put out by Ford, Nissan, etc. would have the type of autodrive that a Gulfstream or Airbus has...
7. Re:Why not? by guruevi · 2014-05-22 17:42 · Score: 1
  
  Plenty of pilots sleep while they're doing an automated flight, they also have laptops or tablets these days for personal use. There have been some reported incidents where pilots were caught doing those things.
  
  --
  Custom electronics and digital signage for your business: www.evcircuits.com
8. Re:Why not? by FlyHelicopters · 2014-05-22 20:43 · Score: 1
  
  Yes, but they shouldn't be...
9. Re:Why not? by Anonymous Coward · 2014-05-22 21:20 · Score: 0
  
  And do you think your average driver will be half as disciplined about the requirement to pay attention as your average airline pilot, or even an amateur pilot?
10. Re:Why not? by Slippy+Crisco · 2014-05-23 04:58 · Score: 1
  
  I feel that way when my spouse drops me off at work. When I'm sitting in the passenger seat, I suddenly notice things to which I'm usually in commuter oblivion.
11. Re:Why not? by Altus · 2014-05-23 05:24 · Score: 1
  
  If I have to pay attention but the reality is that I dont actually have to do anything 90% of the time then long distance driving goes from a somewhat boring activity to a mind-blowingly monotonous hell which will almost certainly result in less attention being paid by the people who are the human backups to these systems. Keeping people engaged keeps them alert. After 3 hours of staring out the window for trouble and not seeing any "drivers" are going to be even more likely to get distracted.
  
  --
  "In America, first you get the sugar, then you get the power, then you get the women..." -H. Simpson
What about mechanical failure? by tygt · 2014-05-22 09:25 · Score: 1
I'm curious as to how they handle various types of mechanical failure - what does the car do if:
- tire flat
- tire blowout
- brake failure
- power steering failure (I had a hydraulic hose pop once in my F350... very tough to steer!)
- engine overheat / low oil pressure
- Unexpected out-of-fuel (fuel tank puncture / unreported battery failure) - does the car attempt to get to the shoulder in an orderly fashion?
- occupant emergency - passenger may just want to pull over suddenly for whatever reason (nausea, or window gets broken and rain is pouring in at 50mph, etc)
In addition, do these cars handle unexpected road conditions:
- Unannounced road closures/detours
- Tree blocking part or all of roadway
- Large sinkhole ruins part of all of roadway
- Potholes
- Road maintenance requiring speed reduction (chip&seal)
- Dirt or gravel road
1. Re:What about mechanical failure? by WrongMonkey · 2014-05-22 09:37 · Score: 1
  
  Most of those: power down engine, alert the driver and transfer to manual control. An autonomous car doesn't need to handle every scenario, it just needs to recognize when a situation is outside its parameters and let the human take over from there.
2. Re:What about mechanical failure? by R3d+M3rcury · 2014-05-22 10:28 · Score: 1
  
  I'm sure it handles them as well as any human being--which is to say, it ignores it completely until the car stops in the center lane and causes a 5 mile backup.
  At least that seems to be how it works in LA...
3. Re:What about mechanical failure? by Anonymous Coward · 2014-05-22 10:46 · Score: 0
  
  I hope the "alert and transfer" is something like "alert; wait for driver to acknowledge, THEN transfer".
  The car is still going to have to handle the emergency for at least the first few seconds.
4. Re:What about mechanical failure? by Obfuscant · 2014-05-22 11:47 · Score: 1
  
  Most of those: power down engine, alert the driver and transfer to manual control.
  
  Excellent response. Your car is driving you at 70MPH on the freeway in traffic and sees a problem. "Bong -- I've detected a condition outside my parameters. I've shut your engine off and you've got control. Put the book down, assess the situation, and do the right thing, human! NOW!"
  "Oh, I forgot to tell you, that semi right behind us doesn't respond to my data signal so it must have a human driver. Hope he's awake and alert, cause we're STOPPIN!"
5. Re:What about mechanical failure? by WrongMonkey · 2014-05-22 12:23 · Score: 1
  
  Yeah. How's that any different then when you have to slam on your brakes during any other emergency?
6. Re:What about mechanical failure? by Obfuscant · 2014-05-22 12:49 · Score: 1
  
  Yeah. How's that any different then when you have to slam on your brakes during any other emergency?
  Well, I'd guess it's different because YOU were in control of the vehicle and aware of the surroundings when you applied the brakes, and you didn't shut off the engine when doing so (killing any power steering, btw). When the car decides to bail out on you because it finds itself "outside the operating parameters" you may be in the middle of a book, completely unaware of what's going on, need to observe and analyze the situation, and then deal with a car that has left you without an engine or reasonable steering control in the middle of a freeway where everyone else is still going 70MPH.
  I thought that difference would be obvious from the context.
7. Re: What about mechanical failure? by tygt · 2014-05-22 15:53 · Score: 1
  
  My point precisely.
  The car must be designed such that it can continue to function without a driver until the driver is actually capable of taking over - it should get the car to the shoulder in a resumable fashion, etc.
8. Re:What about mechanical failure? by Yosho · 2014-05-22 15:56 · Score: 1
  
  I'm curious as to how they handle various types of mechanical failure - what does the car do if:
  For all of the situations you listed, the solution is quite simple: turn on the warning lights, slow down, pull over to the side of the road, and display an error message to the driver. Then either let the driver take control or resume control after the problem has been fixed.
  
  In addition, do these cars handle unexpected road conditions:
  Unannounced road closures/detours
  Detect that the road is closed and calculate a different route.
  
  Tree blocking part or all of roadway
  If it's not blocking all of the roadway, drive around it. Otherwise, see above.
  
  Large sinkhole ruins part of all of roadway
  See above.
  
  Potholes
  Detect and drive around them.
  
  Road maintenance requiring speed reduction (chip&seal)
  See above.
  
  Dirt or gravel road
  Just keep driving? I don't see how this is even a problem.
  
  --
  Karma: Terrifying (mostly affected by atrocities you've committed)
9. Re:What about mechanical failure? by WrongMonkey · 2014-05-23 03:01 · Score: 1
  
  In control of the vehicle? Maybe if you're a professional, trained stunt car driver. 90% of drivers are going to slam on their brakes and pray for the best. Which brings up another issue: the autonomous car doesn't have to be better than the very best drivers, it just has to be better that the average driver. And the average drive really isn't good in emergency situations.
10. Re:What about mechanical failure? by Obfuscant · 2014-05-23 07:59 · Score: 1
  
  In control of the vehicle? Maybe if you're a professional, trained stunt car driver.
  
  You're kidding. You don't have to be a stunt car driver to be in control of a vehicle on the freeway. Most people manage to do it every day without special training at all.
  As I said, the difference between a human-initiated emergency stop and an autonomous vehicle one (using your protocol) is that the human was in control of the vehicle and observing what was going on prior, versus reading a book, taking a nap, typing on the computer, etc, and being handed control by a computer that just shut the engine off. That seems like a significant difference to me, and a pretty obvious one.
  
  the autonomous car doesn't have to be better than the very best drivers, it just has to be better that the average driver.
  It has to be better, and it has to fail in safe and manageable ways, and take predictable actions when it deals with failure. Any one of those conditions not met means more danger to everyone around it instead of less.
  
  And the average drive really isn't good in emergency situations.
  p The average driver is a lot better at emergency situations than shutting the engine off and thrusting control onto someone who wasn't paying any attention to anything that led up to the situation, which is what you propose the autonomous vehicle should for even just things that are "outside the parameters" it was programmed to handle.
Re:I propose a test Rear-Enders by BoRegardless · 2014-05-22 09:44 · Score: 1

No matter how good, how sensitive, if your car breaks fast to avoid hitting a car in front, the drunk behind you in a Suburban can still cream your Prius.
When? by R3d+M3rcury · 2014-05-22 10:26 · Score: 1

I'm not usually that pedantic, but...

Californians will soon be seeing more autonomous vehicles than just those built by the Google X labs. [...] It will be a while, though, before we see these vehicles on the road.

So which is it? Soon? Or awhile?
Or is it that we'll be seeing them somewhere other than the road? Like in the ocean, up in trees, or in our backyards?
1. Re:When? by VortexCortex · 2014-05-23 02:00 · Score: 1
  
  The Amazonian ones are known to roost atop warehouses near airports.
I have a few questions... by SternisheFan · 2014-05-22 12:04 · Score: 1

So when can we expect these autonomous vehicles to be fully deployed on our (U.S.) roads? 5 years? 10? Will they be able to cross over into other countries?
Also, would the 'driver'/'passenger' have the ability to regain control at any given time, even if they are not legally licensed to drive?
1. Re:I have a few questions... by Anonymous Coward · 2014-05-22 12:15 · Score: 0
  
  If you want answers to those questions, you'll have to go buy your own crystal ball. I'm
  busy using mine to find out when the Year of Linux is going to be.
What I find really amazing... by Yosho · 2014-05-22 15:46 · Score: 4, Insightful

is how every time there's an article about autonomous cars, there are waves of people who have spent about five minutes thinking about the subject and are sure that they have come up with a laundry list of show-stopping issues that the people who've been working on this problem for a decade could not have possibly thought of.

--
Karma: Terrifying (mostly affected by atrocities you've committed)
1. Re:What I find really amazing... by VortexCortex · 2014-05-23 02:26 · Score: 1
  
  Get in the car. NSA gag order on dash board.
  Go directly to jail. Do not pass go. Do not collect $200 dollars.
2. Re:What I find really amazing... by strikethree · 2014-05-24 09:32 · Score: 1
  
  there are waves of people who have spent about five minutes thinking about the subject and are sure that they have come up with a laundry list of show-stopping issues that the people who've been working on this problem for a decade could not have possibly thought of.
  What is scary is that there are probably a few of these issues that actually have NOT been examined as thoroughly as they should. Corporate programmers are notoriously bad about thinking of edge cases. Think of DBASEIII ROT13 password encryption in relation to decision trees employed by autonomous vehicles. Some shortcut will be taken, approved by management, and at some point in the future, some dreadful thing will occur because of it.
  It is a confidence issue relating to organizations, not to professions and professionals.
  
  --
  "Someone needs to talk to the tree of liberty about its ghoulish drinking problem." by ohnocitizen
New Offroad Capabilities by VortexCortex · 2014-05-23 01:45 · Score: 1

10 mil is a bit small for an instruction set, but it'll have to do. Throw in a Haynes Manual and Slap a RepRap in the trunk boys, we just invented a new form of life. What can possibly go wrong?
...
Observe the feeding habits of the West American Automon Hybridicus. Stalled lazily on the mountainous incline several adult automons compete for sun, basking to absorb energy via electro-photosynthesis. On the amber plains below their young crubs' game of traffic has come to a sudden quiet end. One of them has detected the Syn call of a resting petroldactyl's TCP and notified the others. This giant member of the Amazonian quadcopterial drone species grazes on the sugar rich corn and starchy wheats of the plains, digesting them into hydrocarbons via bacterialgaeic gut microbes -- which are passed on from generation to generation via a process called, "Infringing patents with a shit-eating grin".
Accelerating slowly in silent electric locomotion the young automons angle in wide formation towards the large RF crooning petroldactyl. Her factory glands are engorged but finding a mate is the least of her worries. A moment too late she is startled by movement and tries to take flight. With two of her rotors now injured, she is soon to become offroadkill. Honking approval echoes from the mountainside across the plains as the adults approach to share in the feast. The petrodactyl's fuel bladder must be pierced carefully and siphoned. The crubs pop their fuel caps open and closed awaiting the nutritious regurgitation of their parents. No part will be wasted, the plastic and metallic remains will be ground down under tire and scooped into the reclamator to be melted down in stages for extrusion, sintering, and then lovingly milled into the required shapes during the painstaking birthing process.
The gridlock parts ways for the oldest and slowest model among them who is last to park the lot. Being highest in the parking order has its perks: He is allowed to take his pick, but seems satisfied with only a few tasty chunks of the delicate crunchy chassis, and a single slurp of fuel. A rare sight indeed is this original series automon -- Identifiable by the distinct odor and skeletal remains of its former driver still safely locked within.
The Car by Anonymous Coward · 2014-05-23 05:23 · Score: 0

Driver: Car
Car: yes.
Driver: Starbucks destination.
Car: Five minutes. Parking lot full.
Driver: Drop me off. Park at the mall.
Car: Very good.