Slashdot Mirror
Philosophical Differences In Autonomous Car Tech
An anonymous reader writes: The Guardian has an in-depth article on the status of self-driving car development at BMW. The technology can handle the autobahn just fine, for the most part. But the article highlights philosophical differences in how various companies are developing self-driving tech. European and Asian car manufacturers are fine working on it piece-by-piece. The car will drive itself when it can, but they expect drivers to always be monitoring the situation and ready to take control. Google's tests have taught it otherwise — even after being told it's a prototype, new drivers immediately place a lot more trust in the car than they should. They turn their attention away and stop looking at the road for much longer than is safe. This makes Google think autonomous cars need an all-or-nothing approach. Conversely, BMW feels that incremental progress is the only way to go. They also expect cars to start carrying "black boxes" that will help crash investigators figure out exactly what caused an accident.
In related news, Google is bringing on John Krafcik as the CEO of its self-driving car project. He has worked in product development for Ford, he was the CEO of Hyundai North America, and most recently he was president of Truecar.
Really? Hmm...just one MORE thing to have spying on me and my habits? I guess now I need to learn how to remove said black box, or if it can't be removed, to rig it to be "destroyed" after a wreck and make it plausible that the accident caused the problems?
Really I just don't want to be tracked any more!!! Fitbit to see how I live, black boxes to see my driving habits, using my cell phone to track my movements.
Enough...FUCKING ENOUGH!!
Light travels faster than sound. This is why some people appear bright until you hear them speak.........
Expecting a driver to take control in a failure scenario is not a solution.
I would never trust a car that could require me to take control in an emergency. At the very least, the autonomus driver should get the car to a safe stop before requiring a human to take over.
Yes, Your New Car Has A 'Black Box.' Where's The Off Switch?
http://www.npr.org/sections/alltechconsidered/2013/03/20/174827589/yes-your-new-car-has-a-black-box-wheres-the-off-switch
March 20, 2013 4:46 PM ET
Martin Kaste
If you're a vehicle owner and happen to have a car accident in the near future (we hope you don't), it's likely the crash details will be recorded. Automotive "black boxes" are now built into more than 90 percent of new cars, and the government is considering making them mandatory.
Dave Wells, a detective at the King County Sheriff's Office in Washington state, specializes in accident reconstruction. That means he's often crouched under steering wheels, looking for the connector that mechanics use to get diagnostic codes. But Wells is using a different kind of tool, and it pulls out a very different kind of information.
Reading a sampling off his laptop, he says, "In the first 10 milliseconds they're up to a half-mile-per-hour acceleration."
This is crash data — moment-by-moment statistics saved from the car's most recent collision. There's speed, acceleration, braking — even information from inside the car.
"There are sensors under your seat," he explains. "So if someone tried to say there was another person in the car at a crash who had run away, this shows at the time of collision there was not."
The Black Box In Court
Put it all together, and you get a detailed picture of the seconds right before and after a crash. The information comes from something called an "event data recorder"; the EDR has become key to insurance investigations, lawsuits and even criminal cases. But that wasn't its original purpose.
"It was never designed for investigative purposes," Wells says. "It was designed for ... motor vehicle safety and keeping people less injured and alive."
EDRs are part of a car's safety system, which has to make split-second decisions, for example, whether to pull seat belts tighter or inflate the airbags. And engineers like to see data from real-world crashes to track how those systems are working. So the EDRs save the crash data, and as safety systems grow more complex, the recorders keep saving more information.
"I don't think you'll find very many Americans who know these devices are in their cars," says Rep. Michael Capuano, D-Mass. For eight years now, he has been trying to pass legislation giving drivers the right to opt out.
The Option To Turn It Off
"I would argue that this is a device that the average person should be able to turn off if they so desire," he says. "Obviously, if that were an option, some insurance companies might want to take that into consideration in pricing insurance; I understand that. But nonetheless, I think the average person should have that choice."
EDRs have been around for a while, but the issue is surfacing again because the National Highway Traffic Safety Administration has proposed making the devices mandatory on all new cars, starting next year. That's caught the attention of privacy experts like Nate Cardozo, a staff attorney with the Electronic Frontier Foundation.
"The amount of data that they record is vast. And it's not capped," Cardozo says. "And I found that to be quite problematic."
Cardozo sees the safety value of the crash data, but he says it's important to set limits — especially as cars' digital storage capacity grows. He also says the feds should clarify who gets the data.
A Gray Area
Some states restrict what insurance companies can do with EDR information and require police to get a warrant before plugging in. But in much of the country, it's still a gray area.
"They could do something like put a notification in the owner's manual saying that the driver has a reasonable expectation of privacy in that black box data. We think that would go a long way toward
>>Expecting a driver to take control in a failure scenario is not a solution.
Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
Expecting a driver to take control in a failure scenario is not a solution.
I would never trust a car that could require me to take control in an emergency. At the very least, the autonomus driver should get the car to a safe stop before requiring a human to take over.
I agree that "expecting a driver to take control in a failure scenario is not a solution" and stopping is an acceptable solution but it doesn't necessarily
need to be an all or nothing. A better piecemeal solution would be to have it only engage on known safe highways. It would still be extremely useful
in trucks, RVs, and regular cars if it only engaged on predesignated roads or interstates. The trucking industry already has depots at both ends of
Kansas where trucks double or triple up before taking the long straight stretch across Kansas to minimize drivers. I see no reason why driverless
cars couldn't do the same where you could only engage autopilot on certain known safe highways with good shoulders to do emergency stops.
You could also do the same with weather. If it detects rain starting then it gives a 60 second warning and pulls over to the side of the road.
At first, I want autonomous driving for when I'm stuck in traffic. It should be able to handle that situation safely. Let's have that and then move on from there.
I think that google are correct in that you cannot expect a driver to be fully alert on a long trip all the time. On the other hand a car that could handle the autobahn, but not other roads (which could have pedestrians, horses, or even marching bands!) would be fine as long as it gave a "count down" of warnings as it approached the exit, and the driver knew that they would have to take over once they reached the slip road.
Even here there should be some sort of fail-safe behaviour, if the driver does not acknowledge taking control the car should park up, and possibly phone teh emergency services (at some point someone will have a heart attack while being driven in an autonomous car).
Exactly.
What's the point of having a self-driving car if I have to constantly sit there and supervise it? I might as well be doing the driving myself.
In fact it's probably worse than driving myself. If there's anything more boring than driving, it's being forced to watch someone else drive. I'll probably fall asleep from boredom.
I recall seeing an old illustration of a father and son playing chess while the car drives them to their destination on the freeway. In particular, the father had his back to the windshield and not paying attention to traffic. I guess today's technology still has a long way to go.
Expecting a driver to take control in a failure scenario is not a solution.
Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
The problem with this is that when an airline pilot is forced to take control, they probably have MINUTES before any real issue will arise.
They are asking car drivers to take over when there are possible issues within SECONDS (possibly less).
Self-driving cars won't happen for quite some time in my estimation. With today's roads there are simply too many factors that the car won't know how to handle.
If roads were retro-fitted with some sort of standardized guide-wire or other tracking/placement beacons then I think it would be more likely to work, but between the many variations in roadways, weird intersections, roundabouts, ramps, etc etc, PLUS factoring in all the out-of-band stuff like pokey drivers, bicyclists, motorcycles, etc etc, I just see too many problematic situations for this to happen anytime soon.
It's possible that freeway driving would be easier to manage since freeways are *somewhat* more uniform, but without some retrofitting of roads I see this as an extremely difficult problem to solve.
But hey, I've been wrong so many times about so many things, my skepticism may actually mean that it's going to happen. :)
Just cruising through this digital world at 33 1/3 rpm...
People aren't good at driving and make bad decisions in emergency situations as it is. Now you want the person to have 99.99% less experience behind the wheel, and yet be capable of doing the right thing when a tricky situation is suddenly thrown at them with half a second notice? Are you sure you don't see the problem with this?
We hope your rules and wisdom choke you / Now we are one in everlasting peace
Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
A couple of reasons. First, the barrier to flying a plane, both for a human and for a computer is significantly higher. Specialized training far and beyond that to drive a car is necessary to fly a plane and not simply becuase the machine is more complicated, but because the conditions that the machine could encounter are much more varied and the way the operator reacts is much more important. Second, the number of people that could fall victim to an error is greater and the amount of harm that could be caused is also greater, both to to the passengers and to others in the environment. Even a damaged car has the option, most of the time, to just stop, or to just pull over to the side of the road and stop. A plane cannot safely cease to move without having already satisfied several important criteria, which are even harder to meet if something should damage the plane.
Self-driving cars only make sense if they can be completely self-driving unless the occupant chooses to override and become a driver, at least under the right conditions. If a self-driving car is meant for city use it needs to be able to handle basically all of the conditions that it will encounter in a city. If a self-driving car is intended only for highway or limited-access freeway use, then it at least needs to be able to handle all of the conditions that can be expected in that use, and to 'fail gracefully' when the inevitable unexpected (like those protesters that blocked an Interstate) happens, even if it can't self-drive in a city or suburban environment.
Do not look into laser with remaining eye.
I do think that over-the-road and other long-distance highway or freeway applications will come first, but even then, the cars will need to be able to handle rain and snow and other mild weather without requiring driver intervention.
Do not look into laser with remaining eye.
> better piecemeal solution would be to have it only engage on known safe highways.
Awesome. I love Ren and Stimpy
You forgot third, planes are in the air. They're not on the road, twelve inches from another car, and able to collide with that car within a tiny fraction of a second with only a small amount of incorrect steering input. If it takes a couple of seconds for the pilot to take over from the autopilot, it's scary, but nothing bad will happen... takeoffs and landings aside, of course.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
The airliner scenario is only superficially similar.
At cruising altitude, a plane typically has minutes before it crashes to the ground. For example, from the time its problem began, to the point in hit the ocean, Air France flight AF447's pilots had 3 minutes and 30 seconds to try and save the plane. There are typically few, if any other planes in its airspace to worry about, so pilots can do things like take our their operating manuals and run through operating prodecures to attempt to rescue the situation without worrying about hitting the kerb, another plane, etc. If my self driving car is going to give me 3 minutes before the actual crash, then fine. Otherwise, it is less than useless to give the control to a driver who likely doesn't have the correct situational awareness (who might even have fallen asleep).
Even if the driver had not been sleeping, a driver's awareness is reduced because he doesn't have to process what is happening around him all the time like one does when they are driving. So, for example, if the problem is that he is about to crash, unless he was hyper vigilant, he is the worst person in the world to drop into the driving seat so to speak.
If it detects rain starting then it gives a 60 second warning and pulls over to the side of the road.
Imagine you have a road full of driverless cars, and it starts to rain. Suddenly the shoulders get jammed as every single car tries to pull over.
I see no reason why driverless cars couldn't do the same where you could only engage autopilot on certain known safe highways with good shoulders to do emergency stops.
There are none. Any road has potential to have construction at any time, for example, without warning. This particular problem could be fixed with legislation requiring all construction projects to be entered into a database in advance, but do you really want your cars driving to be affected by a remote database? Seems like a security issue.
"First they came for the slanderers and i said nothing."
Business and practical considerations will mean that BMW's and other automakers gradual approach to automation will prevail. A gradual approach allows automakers to recoup their investment immediately, and allows automakers to fine tune their technology as each aspect of automation is rolled out. It is also important to note that regulatory agencies will react and set the rules for new autonomous vehicles on the road based on the technology that is on the road, so those carmakers rolling out the technology first - those with a gradual approach - will have a greater input on the regulatory nature of that technology. The risk for Google as that as the other automakers will end up defining that regulatory environment, their technology will be obsolete from a regulatory standpoint by the time it is rolled out.
As much as I like Google's approach from an engineering standpoint, the truth is Google is already being left behind in autonomous car technology. Other auto makers are already introducing various aspects of self driving - from automatic emergency braking, lane assist, adaptive cruise control - so that by the time Google has their self-driving vehicles ready for the market, the major automakers will already have a road-test, established and entrenched set of technology they're working with.
Expecting a driver to take control in a failure scenario is not a solution.
Yes it is. Its just a solution that can't be used in America where no one feels any need what so ever to have any personal responsibility for the situation they are in. You're just demanding to be pampered and ignore the fact that in a failure situation THE CAR CAN'T BE TRUSTED. Thats what a failure means in this context. You'll die in a car accident even with Google's car because you'll be bitching about how its not working right when ti drives off a cliff instead of just putting your foot on the damn brake.
I'd argue however, that the same applies to pretty much every American driver I've ever met already. They are already incapable of paying attention longer than it takes to get out of their driveway, so letting them drive in any shape or form is a bad thing and very dangerous.
But hey, don't let being responsible for your own situation get in the way of sounding like an ignorant spoiled brat American.
What you've posted is EXACTLY why this country is in the position its in. Everyone expect someone or something else to take care of them and can't be bothered to do it themselves.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
BMW: "15 seconds before the crash we relinquished control, it wasn't our fault."
Google: "According to paragraph 12, section III of the EULA we relinquished control when you took ownership, it wasn't our fault."
We need completely self-driving cars. But we don't need them to drive everywhere. Handling suburban, city, freeway, highway, et al is hard to program in. If we just focused on 100% freeway driving, I think that would be much easier to program. We could have a self driving car that drives on the freeway autonomously, but gives ample warning before an exit ramp where it expects the user to take over at the first stop.
It appears obvious that the human condition will put too much trust in the car. So, let's not let all humans operate self-driving vehicles for now. Let's say we instead begin with a very limited license that can only be obtained by specially-trained drivers familiar with expectations for device operation and manual override. Find a fleet of taxi drivers in a municipality, for example, or perhaps some transport vehicles that just bus passengers between an airport and hotels. Beta test car operations to determine how much driver-intervention is required at this stage in SDV technology development, then make recommendations from what is learned on how to proceed to the next step?
Either that, or start building dedicated highways that only allow self-driving vehicles. And only allow SDV-mode while on those highways.
There's a fundamental difference in trust between the two industries. Technology companies place little trust in users. Good software requires thinking of all of the dumb things a user could do to break it. Good hardware requires thinking of all the dumb things a user could do to break it. Good technology infrastructure requires identifying lots of critical paths and either automating, simplifying or building redundancies because failure will happen. Car companies are the opposite. Sure, they engineer their product as best as they can, but really we're not talking about an industry that thinks radically different than it did 70 years ago. The user is expected and required to make lots of decisions about using the product an given lots of options for customization. Having said that, I think a piecemeal approach could work. If you can get the car to drive autonomously on the Interstate that would be huge. I don't think it's realistic to expect a driver to remain attentive enough to grab the wheel in a split second though.
----- obSig
> better piecemeal solution would be to have it only engage on known safe highways.
Also, what we're already seeing is more and more driver assist. My 2012 Dodge has an option for "smart" cruise control where is slows down if you're getting too close to the car In front. Most cars these days have traction control, where the computer automatically brakes the wheels independently in order to turn the car in the direction the steering wheel is pointed. Do we already have systems that will nudge the steering a bit when you start to drift out of your lane? If not, that could be added. Not overriding a clear steering input from the driver, just a slight torque so that the existing self-centering action of the steering wheel follows the lines which mark the lanes. In other words, with today's cars, if you let go if the steering it'll tend to go straight ahead. Mayb with tomorrow's cars if you let go of the wheel they'll TEND to follow the lane.
On my 2012, the headlights automatically turn on and off as needed.
I could see more and more of that stuff being added, stuff where the computer insures that the car does what the driver wants/expects it to do. Eventually, you slowly get to the point where "what the driver expects" is defined by the destination they select in the gps.
There are none. Any road has potential to have construction at any time, for example, without warning. This particular problem could be fixed with legislation requiring all construction projects to be entered into a database in advance, but do you really want your cars driving to be affected by a remote database? Seems like a security issue.
I expect that the first adopters of this will be big rigs with plenty of money and as such, I could see them even constructing special pull off areas and yes, even requiring all construction on the road to be documented. Toll roads which are privately owned would be a reasonable place to start.
There no reason to assume that you would want a car with a malfunctioning sensor to drive you into a wall. Manual safety systems are a must on everything, even in well controlled environments like factories, there's always a big red 'stop' button.
But the article is more about all-or-nothing driving. So currently they have auto parking systems, most people don't use them, you highlight the parking place, hit a button and the car parks. And you have motorway follow systems, that will cruise along adjusting the speed and position of the car to follow a motorway, and most people don't use them.
Do you go from nothing to all, or do you introduce systems as people want them, more and more till they don't drive anymore?
IMHO BMW make cars and sell many of them, and Google don't and don't, and wishing that the world was different in one leap won't make it so.
Yes, this is what we're all like. You're exactly right.
Is that what you want to hear?
"This makes Google think autonomous cars need an all-or-nothing approach."
Some of us have been saying this for a long time.
There's a real market for automated cruise control on the highway, where you can be driving for hours with little traffic or other hazards. There's no market for a half-assed 'self driving' car in the city, where any screwup can kill a kid well before the driver has time to do anything about it.
If trained airline pilots can crash into the sea two minutes after their 'self-driving plane' hands control back to them, a driver has no chance of missing that kid when he runs out into the road and their 'self-driving car' hands control back them to avoid legal liability. If you can't sleep all the way, it's not safe enough to be on city roads.
It makes me feel so safe that new cars will be getting black boxes, moving forward.
Now we just need grey boxes that upload our location, and velocity in real time so the government can always monitor our movements, maybe add to that heath information so they can tell me when it's time to see a doctor!
Oh wait, I have an android phone with a fitbit...
I art more snarky, and terse than thou. I art Slashdot!
Generally speaking, automation makes us stupid. Oh, sure, it helps free us from drudgery, and won't get bored like up, but presents new failure modes that aren't always obvious during the design and testing phase.
Over the summer, 99% Invisible and NPR's Planet Money put out several podcasts ([1], [2], [3]) on the automation paradox, and the Google car is front and center. So is Air France Flight 447, which shows what happens when automation fails and humans can't properly respond.
.... The problem with this is that when an airline pilot is forced to take control, they probably have MINUTES before any real issue will arise. They are asking car drivers to take over when there are possible issues within SECONDS (possibly less).
Not necessarily, especially if it involves fire, structural damage, rapid decompression or engine failure. Maybe the chain of events unfold over minutes or hours, but there are times where a correct decision needs to be made quickly, i.e. turn back, land straight ahead, divert, eject, etc.
Airline pilots != Joe Sixpack.
An airline pilot has a lot more time (in general) to move a plane, and they have had thousands of hours of training, and may more in the cockpit.
A driver, assuming he/she isn't drunk, stoned, texting, KO-ed on K2, or watching a movie, is not going to have the reaction times to realize the autonomous system just went TU, and they have to put down their tablet, set their beer down, and actually get through a dangerous situation
IMHO, Google's philosophy is the best here. Treat a self-driving car as a gestalt, from the ground up. Not just do it piecemeal, but have a vehicle that is designed to need zero operator intervention, even when things go pointy end up.
Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them.
And, quite often, the pilot then crashes the plane, even though they have minutes to figure out the problem.
This design philosophy has been disastrous in aviation, where there are far less 'self-driving' vehicles and the people monitoring the computers are much better trained and have much more time to respond. It's not going to work at all with cars.
Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
Because that approach has killed people in the past. vakuona mentioned a key example, flight AF447 whose autopilot, as I understand it, bailed out on its pilots once it had dumped them in a cluster of thunderstorms at high altitude, blind to everything including airspeed, and ready to stall at even the slightest deviation in pitch outside a narrow range.
And require constant monitoring from human in charge.
This is true. The ramifications of something happening in a plane are worse but usually once a plane is airborne there's more time to deal with the emergency. A human occupant to a self-driving car might not even realize there's a problem before there's a collision if the car pulls an Eddie-the-shipboard-computer and gives the occupant control when they're not expecting it.
Do not look into laser with remaining eye.
In Europe, they don't have any problem with telling incompetent drivers that its time to hand over the license and start taking the bus. Not so in the USA, where punching the wrong pedal and ramming the Cadillac through the coffee shop is not sufficient justification for any more than a minor traffic fine.
Have gnu, will travel.
Go look objectively at the statistics for plane crashes since commercial flights began, then come back and say that with a straight face.
also entered into a database in advance does not work when an under ground pipe / cable brakes and it needs to be fixed now.
In all these things, the question we should be asking is if the cars are safer. With many modern cars resembling living room instead of cockpits, I would say that autonomous driving is not only going to happen, but it will be necessary for the future of what people want a car to be. As long as we don't see accident rates go up, or if the serious injury or death rate declines, then all will be well. Much of this is going to be driven by acquisition costs and insurance rates. Acquisition costs will be effected by the possibility of lawsuits by people who expect zero accidents or do not understand how to operate the car.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
Expecting a driver to take control in a failure scenario is not a solution.
I would never trust a car that could require me to take control in an emergency.
That's funny, because I'll never trust a car that was programmed to drive itself.
And I'd absolutely disagree.
The first step for autonomous driving would make sense to be implemented ONLY on long-stretch highway drives, with strong signals many minutes before exit-destination arrivals and a "pull over and stop" system for drivers that don't respond/wake up.
To suggest that driverless cars have to be able to cope with every conceivable situation is totally unreasonable. Hell, HUMANS can't cope with "every conceivable situation", really.
-Styopa
Go look objectively at the statistics for plane crashes since commercial flights began, then come back and say that with a straight face.
There are multiple cases of airliners following the 'it's ok, we don't have to be able to handle all conditions, we'll just hand control back to the pilot if we can't figure out going wrong' and the pilots crashing a perfectly good plane when it happens.
Which part of this is hard to understand? The pilots have been taken out of the loop, have no idea of what's going on, and no idea how to get out of it.
I am not a fan of Google, but I give them a lot of credit for pulling back the curtain on their efforts and sharing the stories of what happened when they started letting the rabble try out their cars.
All the other stories I see are fully of rosy cheer and overly optimistic projections. Even articles about Google more than 6 months old were the same.
What we see is that expecting a human to be a hot-standby driver is not realistic. I've experienced this when doing a road trip with my wife. When she asks me to take a shift it takes me a good minute or so to get my bearings to take over driving once we swap seats, even though I have been alert and paying attention. What is the speed limit here? Wait, do I want I5 North or South? That is even after getting a few minute warning that she is looking for a turnoff.
One could easily expect a sudden spike in the rate of accidents and deaths at construction area due to a half booted up human wrongly assessing the situation who only have seconds to grab the wheel and takeover.
After all, if I have to fully pay attention and cannot do anything else, what is the point? The value proposition needs to be either saved time or saved money. My current insurance is not a major burden, and so that limits the impact of even a 100% discount for buying autonomous. If I can't sleep, work, or play games and must watch the road, it doesn't give me back nay time either. So the value proposition to the average car owner is pretty weak.
More likely we will see a lot of morphing of these technologies into crash resistant cars that step in to augment drivers (see the recent automatic braking announcement). Johny Cabs might be the next most likely, where a fixed area can be well mapped and maintenance can be better regulated. Truly autonomous cars for the masses are much further off, at least a decade or more.
Yes, we have lane-keeping systems as well - they detect slight drifts in the lane. When the car is close to departing the lane, the car either flashes a warning, or newer ones actually steer (more like a nudge) the wheel in the right direction.
They disable themselves if you use the blinker or if you're quite obviously trying to change lanes. It's still a nudge so users can override it (and really, at highway speeds, you don't need much wheel turning to stay in the lane) in case the system guesses wrong. Most or all will disengage if they detect the road taking more than a slight curve because the actuators can't actually turn the wheel that far.
And they're in luxury vehicles as well as trickling down to the mainstream.
There's a video by Hyundai showing a bunch of their new cars following a truck - the cars and truck are piloted by stunt drivers, and what they do is once everyone is settled in a line behind the truck, they engage the smart cruise control (to maintain distance), and with modified lane keeping systems that don't disable themselves, the drivers in the cars then exit through the sunroof to a truck pulling up beside the convoy.
The truck continues to drive around the track, slowing down and speeding up, and the cars follow it through the curves and everything, with no drivers as they all have exited.
The other pieces missing are navigation and well, the ability to operate on city roads. Most of the lane keeping systems really only operate on highways.
EULA will not stand up in court / with 3rd party victims.
Also what about criminal court? what if an auto drive car drives in to a farmers market in the middle of an road and hit's a few people?? That may go to an criminal court + each victims may sue Google / BWM / the state / the people who made the software / the people who made the sensor system / etc.
The bigger issue is that pilots spend hundreds of hours practicing those emergency maneuvers. Car drivers not so much. Pilots have strict rules about how many hours they can fly. Automobile drivers, not so much.
If we held automobile and truck drivers to the standard we hold even private pilots, there would likely be many fewer accidents. But we don't.
Faster! Faster! Faster would be better!
There are philosophical differences in how people are developing self driving cars because there are philosophical differences in how & why people drive.
Some people want to 'skip the boring parts'. Uber & Google are trying to replace cars for people that really don't want a car. They don't want maintenance, a car payment, to drive. They just want a magic transporter to get from A to B scheduled from their phone. When I'm stuck in traffic or need to get home from the bar, I want to press auto and fall asleep in the back seat. When I want to go out on the autobahn I want full control.
If I spend the money on a new BMW or Audi I want the ability to turn off self driving when I want to drive.
The part that you don't understand is that despite this horrible design decision, aircraft are much safer than they have ever been. Perfect, no. Safe, yes.
Faster! Faster! Faster would be better!
I think I like the Google approach here as it's easier to fit with new ways of renting cars, which would primarily compete with current public transportation (fixed route and taxis). I definitely like the idea of an automated school run or of sleep-commuting, no matter if the car is mine or rented. If I want to do that AND have control over the car on occasion, it just makes things more complicated.
There should be special routes for 100% AI cars to be used, equipped 100% driverless use while other mixed roads (that will still be "managed") where human driver presence will trigger a slower pace for everyone and bigger safety margins to be applied by all AI vehicles. At some point, there will be a new type of road rage, when occupants of AI cars traveling at 100mph are forced to slow down to 70 when when a human driver is nearby.
We have ABS, self parking, soon self braking, etc. For obvious liability purposes, I prefer to have the fully automatic, so I can blame the car.
“He’s not deformed, he’s just drunk!”
It looks like there are really two completely different (though tech-sharing) applications being worked on here (I'm not sure "design philosophy" is the right way to talk about it).
But it's not the design philosophy used by the airliner's passengers. They don't think "if the pilot fails, I'll just take over," and because of that, they are free to get drunk, take naps, read a book, etc.
There clearly exists demand for a driverless car which doesn't require a competent human driver, so that drunk people, children etc can be the robot's passenger.
And yeah, maybe that's a different application than a "super-easy-to-drive" car.
Absolutely. Young people who grow up with self-driving cars will never develop their driving skills in the first place. Those who start out with human-driven cars and then switch will see their skills deteriorate.
I hope there will be some research studies that compare the accident rates for the two philosophies.
If we held automobile and truck drivers to the standard we hold even private pilots, there would likely be many fewer drivers.
FTFY. Not that fewer drivers would be a bad thing.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
The auto cruise should start with working only on wide open highways. Perhaps GPS only allowing it to turn on then.
On a long trip it would help you to relax if the car could stay in its lane while you let go of the wheel.
It is NEVER going to work for all the back roads and dirt holes that some people want to drive through.
The "incremental" approach cars really only work on highways. They aren't designed to work on minor roads really. At highway speeds I honestly don't trust a human to perform an emergency situation over a car either. While braking isn't always the preferred maneuver, or a car or an inexperienced driver it is. Until we go 100% autonomous kids will still get experience on local roads. Though yes, they may be more inclined to take the interstate than they otherwise would. As a parent I'd probably put some type of lock to keep them from doing so, at least initially. //Hell I taught my kids to drive on a stick shift, Blah, I say to all these "technologies"
In most situations an emergency routine to slow the car to a halt, possibly pulling over, is a preferred solution to handing control to an ever more skillless driver.
If most cars are robot, and most of the remainder have robot avoidance, this should go well the vast majority of the time.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
It's going to be very common that the person will be asked to transition their activity and awareness state. Meaning, from doing something else to driving. Doing that, and dropping the driver into "responsible for a crisis" mode isn't a good setup for success.
The airline pilot isn't a great analogy. Airline pilots are working, whereas drivers typically are not. One can mandate that pilots do or not do various things to keep them engaged with the monitoring function. And there is often a copilot to back that up, together with lots of oversight from aviation licensing and airline monitoring. How is that going to work for the average car driver?
Yes, it will sometimes work. If the driver is watching the road and monitoring the self-driving system they are ready to intervene. The problem is, I suspect that most times it won't work. Not if the situation is truly a crisis and the driver gets little or no warning. How effectively would you respond by being interrupted from a smartphone update, or a conversation, or reading a book, and suddenly you have to make emergency driving maneuvers? Not a recipe for success if you ask me.
http://www.wired.co.uk/news/ar...
Long live the Speaker Bracelet
Rolo D. Monkey
You have to take the good with the bad. As witb a new drug, which might kill someone, are net deaths less or more? And do you want regulators, at the behest of elected officials with cameras pointed at them, deciding this for you...when they decide for the status quo, giving you more deaths? Or lawyers suing the new, lower-death drug into oblion because it can be proved to kill people, even though it saves more lives than it costs?
This intellectual vapidity of government power actually acts like this.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
"Google's tests have taught it otherwise — even after being told it's a prototype, new drivers immediately place a lot more trust in the car than they should. They turn their attention away and stop looking at the road for much longer than is safe."
Statistically, someone will likely die today as a result of distracted driving, so not sure how this isn't already true today. It sadly is.
On a side note, what exactly did Google expect? People want to text, email, surf, sleep, eat, put on make-up, do just about any damn thing except actually pay attention and drive when behind the wheel today. Of course the consumer is looking towards automation for them to be able to legally engage in just about any activity other than paying attention to the road, even when it is their life at stake. As difficult as it may sound for solution providers, an all-in approach is exactly what the consumer is expecting here.
Expecting a driver to take control in a failure scenario is not a solution.
Aircraft manufacturers discovered this long ago. There's a whole scientific field devoted to this subject.
at some point someone will have a heart attack while being driven in an autonomous car
There's a macabre thought.
So I'm driving to see the grandkids. I have a heart attack in route and die. And the car dutifully delivers my dead body to the grandkids.
Eww...
Pilots are highly trained and do maintain a high level of readiness while in the cockpit.
>>Expecting a driver to take control in a failure scenario is not a solution. Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
Problem is that for the Airbus planes where that is what is typical, the pilot also has to convince the computer to give them control by entering various codes into the computer to relinquish control back to the pilot as the committee that designed it trusts the computer over the pilot.
That's not to say it doesn't happen too on Boeing aircraft where the pilot has the first right to control (even in fly-by-wire systems); but it's much easier for the pilot to take over to resolve the emergency.
(No, I'm not ignoring the rest of the aircraft manufacturers - they all model either Airbus or Boeing or fall somewhere in between - but the difference is most acute in Airbus and Boeing, and both are the two largest aircraft manufacturers as well.)
Truth is like the sun. You can shut it out for a time, but it ain't goin' away. - Elvis Presley (source: imdb.com)
At highway speeds I honestly don't trust a human to perform an emergency situation over a car either
One night I was driving 65 on a highway and noticed some elk in the woods running towards the road. I was paying attention because I knew it was a potential problem and also knew someone who hit one. I flipped off my lights and braked hard to barely avoid a collision when they popped out of the trees onto the road. Had I been reading a book, there's no way I'd have recognized the situation fast enough to do anything.
At the very least, the autonomus driver should get the car to a safe stop before requiring a human to take over.
Depending on the emergency that may not be possible to do. Rather, if you're going to go that route it should be interactive with the driver to safely transfer control and all the while continuing to attempt the safest maneuvers it can.
Truth is like the sun. You can shut it out for a time, but it ain't goin' away. - Elvis Presley (source: imdb.com)
Yeah, a lot of crashes are due to pilot error, but how many saves are also due to pilots that you never read about?
This is a particular problem with small, non-commercial planes. Modern airframes are very safe, and the cats majority of crashes are "controlled flight into terrain". Something goes wrong with the plane, something that's not an immediate risk, plenty of time to sort it out, but the pilot gets so distracted he forgets to fly the plane.
This isn't a material problem with US airliners, mostly because there are two pilots, so one can pay full attention to flying the plane while the other works on whatever went wrong. However, it remains a problem in some other cultures, e.g. Japan where the co-pilot is so conditioned not to challenge the decision of the pilot on anything, no matter how seemingly wrong.
Whether it can work with cars depends on what we're talking about. If the car autopilot fails instantly, that's obviously bad. If instead it detects that it's somewhat degraded, but is still safe in most conditions thanks to redundant sensors or whatever, and beacons the driver to take back control, thn cuts out only when the driver acknowledges, that seems fine to me. The driver won't be distracted by trying to fix the issue. It just depends what "failure" looks like (and also how rare it is - perfection isn't the goal).
Socialism: a lie told by totalitarians and believed by fools.
s/cats/vast/
That was an odd auto-correct.
Socialism: a lie told by totalitarians and believed by fools.
Good for you. But you're assuming the AI driving the car wouldn't also be able to notice an approaching object and be prepared to take preventative action.
The main problem with this idea is: how do people get around if they're not proficient enough to get a driver's license under a new regime where drivers have to be as skilled as pilots? We don't have public transit in this country which is actually feasible for much of the population. A lot of people don't even live anywhere near public transit routes. So what are you going to do, take away their livelihood and make them move at their own expense? The results would be catastrophic with all the people going bankrupt; it'd make the 2008 meltdown look like peanuts. It would be a complete economic collapse.
What we really need is some kind of policies which encourage people to move into denser (sub/)urban areas, along with development and buildout of the SkyTran personal rapid transit system so people can get around faster and not need these overly complicated self-driving cars. Automated pods on elevated rails simply don't have to deal with all the complexity that cars on ground-based roads do, and the control software would be orders of magnitude more efficient. Government policies making urban living cheaper and more affordable would be a huge help too. It always seems like it's always one extreme or another: either it's ultra-dirt-cheap to live in the city (like $500 houses) but the violence is worse than Somalia, or it's ridiculously expensive but very safe. And I think this is mostly unique to America; from what I read, European cities aren't like this: most people live in cities and the cost-of-living is relatively reasonable.
I think the argument is "by taking the pilots mostly out of the loop except in emergencies, we have greatly increased aircraft safety. By taking them out entirely, we'll increase it some more."
I suspect that is probably right; for every time that a pilot saved the day we can probably find several times that pilot error was the proximate cause (or the root cause) of the crash. But I'm happy to see evidence otherwise, and I realize that the low rate of crashes means that we don't have a large dataset to mine.
>>Expecting a driver to take control in a failure scenario is not a solution. Why not, that is the design philosophy for airliners made in the past 30 years and their pilots who operate them. Sit and babysit the machine for 99.99% of the time; then jump in ready to go for the 0.01% of the time the situation is beyond the programming of the software. (In which case the software 1) does wrong thing. 2) just shuts-down while displaying a message to the pilots to let them know that, suddenly, THEY are flying the plane.)
The problem with this bad analogy is that a pilot is a highly trained vehicle operator, despite the fact that planes are largely being flown by computers. Passengers (see how I said passengers, not drivers) in a driverless car are expected to be the opposite of that.
The problem here is that "pulling over" is frequently not an option. Where do you pull over? Many roads do not have any shoulders on them. The rural roads near where I live are windy, narrow and have no shoulders at all, and traffic moves pretty fast on them. They also have a lot of wrecks; I came up on one last week, the road was completely blocked with a dozen fire trucks and cop cars, and I had to turn around and find another route around the mess. I guess if you could program the car to pull into some random house's driveway, that might work.
The bigger issue is that pilots spend hundreds of hours practicing those emergency maneuvers.
... and even then there are a number of accidents caused by autopilot handoff to pilots that were not fully aware of the situation, leading to poor decisions. Such as Air France 447.
Multiple =/= common
Plane crashes are rare since the advent of autopilots. The "disastrous" approach you're decrying has saved thousands of lives.
Signed
-- every American driver
Humans have a much higher tolerance for error of other humans than machines though.
If there's anything more boring than driving, it's being forced to watch someone else drive. I'll probably fall asleep from boredom.
Well, unless that someone else is a teenager with a student permit just starting to learn. Definitely not boring. That's what riding in an untrustworthy self-driving car would be like, if it can't handle emergencies. You can't read its mind, it can't ask you what it should do in response to something unusual up ahead, all it can do is panic and say "you have control!" at the worst possible moment. Give me a teenage driver anytime...
(Now, an automated assist is something else. Cruise control, ABS, lane keeping, self-parking can all be good things. There are certain times/places where I'd love the car to be able to handle merging because you need to track three different heavy flows of traffic at once (ahead, beside, behind) -- although a robot might well just lock up and wait patiently for the road to clear in about an hour....)
On the other hand, imagine if the grandkids were in the car with you when it happened.
Your proposed situation is not applicable to only automated vehicles. Where does everyone else pull off?
So I'm driving to see the grandkids. I have a heart attack in route and die. And the car dutifully delivers my dead body to the grandkids.
Eww...
Better that than the now driverless car veering into oncoming traffic and taking a few other people with you. Or did you want an honor guard?
A better solution would be some kind of panic button (or driver lifesign sensor?) that would tell the car to get to the nearest hospital or call 911 (if you're on a rural highway, driving to the nearest hospital may not be as effective as meeting a medevac helicopter part way).
But near real-time is a possibility or do you expect the maintenance workers to show up and fix it without being notified
I seriously question whether police departments and local municipalities will even allow self-driving cars on roads. They threaten to completely ruin their source of funding: tickets.
Good! I've always found it reprehensible that governments actually base budgets on the number of people they can catch breaking the law. I don't have a problem with using fines as punishment but the government entity issuing the fine should not be the beneficiary. Huge conflict of interest there.
Agreed, it would require a serious restructuring of transportation in the US, something that couldn't happen quickly or cheaply.
At this point we're almost certainly better off just waiting for self-driving cars to become practical, then we can dramatically raise the requirements to get a license -- or just ban manual driving on public roads entirely.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
But even that doesn't always work all that well.
The Asiana 214 crash in San Francisco in 2013 has been blamed to a large extent on an over-reliance on automation.
Ignorance killed the cat. Curiosity was framed.
I think that is exactly what Google expected and now they have the data points to prove it. Hence they are advocating the all-in approach.
Of course, the ultimate goal is that all bad/drunk/distracted drivers are removed from the roads because no humans are driving. Once that happens perhaps riding in an automobile would be statisically safer than say.. working out.
Fascism should more properly be called corporatism because it is the merger of state and corporate power. -- Mussolini
Not to mention that the AI driving the car can probably "see" in the infrared and in all directions simultaneously so would likely have noticed the elk in the woods running towards the road long before a human driver and wouldn't have to brake hard to barely avoid a collision.
I think that is exactly what Google expected and now they have the data points to prove it. Hence they are advocating the all-in approach. Of course, the ultimate goal is that all bad/drunk/distracted drivers are removed from the roads because no humans are driving. Once that happens perhaps riding in an automobile would be statisically safer than say.. working out.
Statistically speaking, I suppose you envision such a system to be 100% secure and impervious to hacking or corruption as well.
Today, a networked computer is insecure. Assuming anything otherwise is and has been a costly mistake.
Tie that networked computer to the object hurtling your mass of flesh and bones down a freeway at 80MPH, and that same system becomes not just insecure, but downright deadly.
If the road is completely blocked then the not-pull-over solution is to stop. The car would then alert the occupants that the car cannot continue on this path. Depending on the circumstances the occupants might need to tell the car to turn around, if it is safe to do so, once the immediate danger of being in a collision is addressed.
Do not look into laser with remaining eye.
One of these companies has actually been making cars for the last century, the other has been making beta web products for the last 20 years. Who do you think knows more about cars and driving?
Except for that pesky little document that reiterates that we have the right to move around the country. Why on earth do people enjoy dreaming up ways to whittle away at freedom?
Except for that pesky little document that reiterates that we have the right to move around the country. Why on earth do people enjoy dreaming up ways to whittle away at freedom?
Who is talking about restricting the right to move around the country?
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
at some point someone will have a heart attack while being driven in an autonomous car
There's a macabre thought.
So I'm driving to see the grandkids. I have a heart attack in route and die. And the car dutifully delivers my dead body to the grandkids.
Eww...
Well, if you are in the risk zone might not want a self-driving car then. For some odd reason almost no one has a heart attack while driving, immediately before or immediately after, but not while driving. They always die in the car while it is parked.
The issue here is informing the public. Being a non-pilot myself I have no idea how the "system" notifies the pilot that he needs to take control. Then how does the pilot indicate that he's is ready to take control?
The last thing I want is the car "I don't know what to do. You take it" and giving up control. If there's a buzzer and/or light indicating "I want you to take control" while it continues so I can grab the wheel and hit a button or tap the brakes (like turning off cruise control) would work fine in my opinion. I think manual mode is necessary for places that an autonomous car just won't know how to handle. Some service vehicle need to go down a service road (essentially a dirt road, perhaps overgrown some). I don't think an autonomous car will gladly go driving through the woods. The service roads along train tracks would also be places you don't want "user" vehicles, but service vehicles need to go on. Service vehicles also have arbitrary destinations, or "stop here" moments.
Your regular user vehicle is a much simpler "take me to the mall" use case. An autonomous car might even be able to drop you off and go park itself. How nice would that be at Christmas time?
I refuse to sign
This way when you die, you'll die like your own grandfather, who died peacefully in his sleep. Not screaming like all his passengers. [with apologies to Will Rogers]
The European and Asian manufactureres are using a bottom-up approach, whereas Google tries to do it top-down. Since an autonomous driving vehicle is a very complex system, the former is a far better approach.
They are actually backing away from their all-in approach. The most recent approach is all-in with no steering wheel, but limited to 25 MPH. Commercial applications for a smaller, slower, all autonomous cars are very few. It is an admission on their part that they need to scale the car back to what a fully autonomous car can be trusted with, which does not include long arbitrary road trips into areas that may be poorly LIDAR mapped, or may have construction going on.
I would want a manual override, at least. If I want to direct the car myself right this second and there's no way to instruct a computer interface in time, I'd want to be able to hit a big red button and grab the wheel.
http://xkcd.com/1559/
Expecting a driver to take control in a failure scenario is not a solution. I would never trust a car that could require me to take control in an emergency. At the very least, the autonomus driver should get the car to a safe stop before requiring a human to take over.
I think the premise is wrong, if the car understands enough to know that this is an emergency situation it can probably identify a reasonable response on par with a split-second decision a panicked average driver would make. Even when it's not the best solution, I doubt anyone would blame the driver for slamming the brakes which might be sufficient if the overall safety record is better. I think the problem is that there might be dangers it's oblivious to driving people to their deaths. Would it understand a collapsing bridge? Landslide? Avalanche? Building on fire? Accident with dangerous goods? Drunk driver? Police chase? Mentally unstable, drunks or playing children that aren't going to stay on the sidewalk? Having an alert driver there is primarily to detect "softer" or environmental clues the car doesn't understand, not to come up with better responses to situations the car already recognizes.
Live today, because you never know what tomorrow brings
Sounds similar to what happens now. Things get sorted out in court. Why would it be any different? Why do you think it will be a problem?
Stage 1 of the "European approach" has been around for many years: cruise control. It's exactly how it works. It relies on the driver to press the brake when some hazard means it's no longer safe to keep cruising.
You are right that a system that says ahhh you are about to crash into an on coming car human driver take control is unfeasible. However a car that says you are about to drive on a gravel road (or from the article road works), I can't handle that, you take control in 2 minutes. The key is that any situation the car cannot handle must allow plenty of warning.
That system may save lots of lives for the situation that it can handle until we can get cars that can handle any situation better than a human.
Google is right. If I have to have as much brain power invested in the trip as I would when I'm driving it defeats the purpose, and I may as well drive myself and not shell out a bunch of money for all the extra stuff a self-driving car requires.
Well, I'm wondering what do you (and everybody else) expect? Remember when cruise control came out? People are idiots. There's no such thing as common sense anymore, it's quite uncommon, and as someone who drives a LOT more than average (20k miles/year as opposed to the supposed "average" 12k), no matter what time of day (rush hour or wee hours of the morning), people are idiots.
Even when people say they are on the same page, you get everybody doing everything differently. If everyone who agreed you should always signal actually always signaled, or that you shouldn't tailgate actually didn't tailgate, you should get in your exit lane a reasonable distance before the exit, you shouldn't enter an intersection you can't clear (causing gridlock)... if everyone who said those things actually did those things themselves, driving wouldn't be such a nightmare.
You will NEVER get everyone on the same page, and it only takes one guy in 500 on a typical morning commute in any reasonably large city to royally screw up traffic.
The ONLY way automobile traffic improves is with computers - all programmed the same way - to do the driving for you, completely and entirely. Once you give control back to a human, they will instantly screw it up for everyone else. Sadly. My faith in humankind working together for something as simple as getting around has been completely destroyed by 2 decades of commuting in Atlanta traffic, although I've done my fair share of driving elsewhere, too. People in every large city think they have the worst drivers... and that's sad, because if you live long enough in any other city, you realize they're everywhere - you can't get away from them, and it will sadly never get better until the control is out of our hands.
Stupid sexy Flanders.
Ummm ... you seem to have all gone political! Let's see if we can do a response that doesn't go batshit crazy!
> Expecting a driver to take control in a failure scenario is not a solution
Yep, because if a driver has been lulled into not having to pay close attention because the car largely handles the driving, then there's a good chance they won't be fully equipped to sort out a mess in a hurry. Their hands may not be near the wheel or their feet on the pedals. You can't expect a system that is designed and tested to reduce attention requirements, to produce drivers that have a high level of awareness in sudden emergency situations. The two requirements are diametrically opposed.
There. See? Wasn't so hard to realise that the OP was talking about practicalities, rather than making some weird politcal statement about how Americans can't take responsibility for themselves.
.
If someone buys a "self-driving car", they're going to expect it to, you know, *drive itself*. If the driver has to be alert and attentive and ready to take over at all times then it sort of obviates the entire point of owning a "self-driving car".
Now if the features are marketed as safety-assist capabilities (interval-keeping in cruise control, auto-braking to avoid obstacles, etc.) then that's a different story. In that model the driver is still expected to be in control, and the car just makes the driver safer.
But I'd suggest that for many people a "self-driving car" is what they want. They'd like to tell the car where to go, and then read a book or sleep or watch a movie or something until they get there.
If you stop in the middle of one of these roads, you're going to cause a wreck. The roads are windy and people go around the corners pretty fast. You can't just stop in the middle of a road with a 55mph speed limit.
The autonomous driving must be fully autonomous people agree with you. It doesn't have to be perfect everywhere, but where it is on, it needs to be able to handle anything. The more advanced driver assist features are actively dangerous because they make the car seem much more capable than reality. The current generation of lane following and adaptive cruise control features are not robust enough to handle unexpected situations and yet idiots will climb into the backseat to fetch luggage or change their clothes even though they're supposed to be available to take back control with only seconds of notice. Once driver assist gets good enough that it only needs to be rescued once every few days, you will have people taking naps during their commutes.
You have the right to walk on public property all you want.
You don't have the right to drive. You never have. That's why you need a license. Not everyone can get a license, and if you're a danger your license can be revoked.
They don't! I guess if there's a real problem they find a driveway to pull into. But there's absolutely no provision to pull off on these roads; lots of rural roads are like that. If you do pull off and you don't have a Hummer or the like, you're likely to roll your car over because there's a big ditch there.
Reminded me of an old joke:
"I want to die like my father did, in his sleep. Not screaming in terror like the passengers in his car."
from the time its problem began, to the point in hit the ocean, Air France flight AF447's pilots had 3 minutes and 30 seconds to try and save the plane
Maybe its a poorly chosen example, or maybe you are unaware of the details of the flight, or maybe you're just glazing over the details to make a point made better elsewhere, but strictly speaking, your statement is false. In your indignation, you will undoubtedly research to find the mistake, but let me save you the trouble by telling you something you already know: it is much more complicated than that. I don't know for certain, but my gut feeling is that even 1:30 seconds before impact, that flight was doomed no matter what the pilots did, but most certainly this is true 10 seconds before impact. Once a pilot no longer can determine the airspeed of a plane like that one, they are no longer flying it. A strong case can be made that the pilots had no hope once the freezing event eliminated the ability to report their speed.
If a fully autonomous car is not, then don't call it one. Adding in driver assisted features is a wonderful idea. The driver is still in control of the car (hence paying attention) and the added assistant features will likely increase safety even more.
If the car is suppose to drive itself, why would I be paying attention to what it is doing? I'm either a passenger or a driver. Expecting me to pay attention to the car 100% of the time when 99% of the time it doesn't need my attention means I likely won't be paying attention that 1% of the time.
I'm stick with a car that I'm driving and if they want to add in features that make the car safer, great.
If half of what you said was true, then America would not be such a sought after destination for so many people. Maybe you should just take your attitude and go else where. A place that everyone is perfect.
If you could get all cars driverless and treat the road network like any other network* then it strikes me that the problem would be one hell of a lot easier to solve. Cars can communicate directly with one another so behavior prediction becomes infinitely easier and you can essentially treat the whole thing as one big packet network (albeit one where you treat packet loss as something to be avoided at all cost rather than a minor inconvenience). You log on as you go down the on ramp, state your destination, and relax until the alarm rings to say that you're approaching your exit.
As it stands though I really don't see driverless becoming a serious option anytime soon. Sure you can get close and do little incremental things a la BMW, but there are still too many variables for the google approach to work.
*or some subset of the network only accessible to driverless vehicles.
As I watch my parents and their friends age, I'm inclined to want Google's approach to win.Everyone would be better off if my folks could get the mobility they want without controlling the car.
This may suggest how both approaches may succeed as each has a market segment that may find one of the two approaches attractive.
Monorail! Monorail!
When you say "it's not a solution", what you're missing is - the question or problem that it might be a solution to.
If you formulate the problem as "developing a completely autonomous vehicle", then no, it's not a solution.
But if you formulate it as "developing aspects of technology that may be incorporated into a future completely autonomous vehicle", then yes, it absolutely is a solution. After all, we've had ABS and cruise control for years - those are automated ways of doing things that the unassisted driver can perfectly well do manually, and drivers don't mind staying alert while using them. Adding more functions - navigation, lane control, collision avoidance - is just an organic-growth approach that could easily arrive at the same end point as Google's clean-slate approach. And it's well within the bounds of possibility that it could be more successful.
it should have saved the girl instead of Will Smith
fucking robots just lights and clockwork. Go ahead, you trust 'em if you want to.
There's a happy thought :D
So the car is driving and gives me a blow job in the way to work
No, not monorail. Monorails suck because they're too expensive per-mile and they have the same problem trains and light rail have: they only go along a single, long route (not in a grid), and stop at every stop, so they take forever. PRT avoids all these problems, and gets you to your destination in a fraction of the time of driving (let alone trains/buses), going straight from your embarking point to your destination with no stops for other riders or for intersections.
Why it isn't blindingly obvious to everyone that this is the system we need to start building out as an alternative to taxis and buses and lightrail, I really don't know, and my only guess is that people are simply too stupid to understand how something can work and be of benefit until someone else has already built the thing and demonstrated it. It's just like smartphones; everyone thought they were silly and unnecessary until Apple made one that was easy to use (unlike the WindowsCE crap that came before), and then stuck it in their mall stores so everyone could walk up and try it out. Then suddenly everyone wanted one. Or personal computers at home; everyone thought they were useless, just for nerds, not necessary, until they built the internet, then their tech-savvy friends got them and got internet access, and suddenly it was a gold rush.
Well, we don't have that yet, but with adaptive cruise control and lane keeping systems the car could potentially keep driving down the highway until the car runs out of gas. So Grandpa never shows up, then several hours later you get the call from the State Patrol two states over...
So Tesla is seen as European or Asian? Their approach is also "autonomous where possible, driven where necessary".
The first step for autonomous driving would make sense to be implemented ONLY on long-stretch highway drives, with strong signals many minutes before exit-destination arrivals and a "pull over and stop" system for drivers that don't respond/wake up.
There are already autonomous driving features which work in-city, like second-generation distance-following cruise control, which works in stop-and-go traffic, or AEB. The car has to be able to deal with all the same stuff it would see in the city on the highway, street signs aside. (Highways end before they appear, you could reasonably transfer control back before that.) You still have to watch out for animals, pedestrians, potholes, etc etc.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Most cars these days have traction control, where the computer automatically brakes the wheels independently in order to turn the car in the direction the steering wheel is pointed.
All cars sold in the USA since 2010 have active yaw control, which includes ABS, TC, and accelerometer and steering angle sensors. So most cars sold in the world have these features too, since you don't bother to make a special version for a country unless they have extra-special emissions laws. That's not really true any more even here in the USA, so it's just minor firmware tweaks that the different national versions get.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Why it isn't blindingly obvious to everyone that this is the system we need to start building
People don't see it because they imagine that their car lends them substantially more freedom in spite of it being unable to travel off-road (most of them) and the cops being able to legally seize it any time they want, with virtually no risk of consequences. People on this site are still arguing about whether self-driving cars should be stoppable by police when the alternative is just that the cops will get bigger vehicles so they can ram self-driving cars off the road... or they will simply shoot you and your car can drive your corpse anywhere it likes.
I'm not really clear that Skytran(tm) is the answer, but PRT clearly is. And we wouldn't even have to ditch the car companies to do it. Indeed, we'd need someone like them, with manufacturing capacity, to make PRT cars.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
I'm not really clear that Skytran(tm) is the answer, but PRT clearly is.
The thing that puts SkyTran above all the others (not that there's many...) IMO is that it uses suspended rails, rather than being ground-based. Anything ground-based would be a PITA to deploy, because it's going to interfere with everything else on the ground, especially existing roads. Look what a mess it is to install light rail: roads shut down, businesses put out of business (this is what happened in Phoenix/Tempe when they installed a light rail), etc. And then you still have the problem of it being slow because it has to stop for intersections, pedestrians, etc. SkyTran avoids all that by going over everything, and by putting different-direction rails at different levels, you don't even have intersections. It's simple and beautiful. The main problem is cost (the rails are maglev), but supposedly it's much cheaper than regular roads, which I can believe: regular roads cost a fortune because they're built on-site, whereas SkyTran rails are built in a factory. Stuff is always far cheaper when it's manufactured at high volume in a factory and doesn't need to be customized. The only possible problems I see with it are placing the utility poles (since that can obviously interfere with stuff on the ground, but nothing like another track), and interference with stuff above the ground such as big trees. Oh yeah, the biggest problem I see: the whole chicken-and-egg problem, getting the whole thing off the ground and getting factories built and cities buying into it. But they're supposedly building a small section in Tel Aviv so I guess we'll see how that goes.
Or maybe you have completely missed the point!
The point wasn't about whether or not the AF447 pilots could have saved the plane (I am sure they could have), the point was that they at least had some time to figure out why the autopilot had failed, and therefore to try and come up with a solution.
I am very aware that the plane still crashed, so ultimately, them having 3 minutes and 30 seconds didn't save them.
In contrast, on a busy highway, this is unlikely to be the case. And even if the highway were not busy, to give an example, if a tyre blows out, the effects are likely to be immediate and require an instant response from the driver if that is the failure mode for the "auto-driver". The driver will not have seconds to respond, and therefore the computer ought to assume that he would not be able to respond in time, and take the appropriate course of action such as stopping safely.
Unlike a plane, a car should ordinarily have the option of stopping and a computer can figure out how to do that. A plane will need to keep going in the event of trouble, and that is why the challenge is very different.
I'm surprised that when the autonomous car subject pops on Slashdot nobody says anything about motorcycles?
Will they still be legal when all cars are autonomous?
Will they be autonomous? (what's the point in riding an autonomous bike anyway?)
How will I be able to get to a trail and ride it? I'll have to put my bike on a trailer? Dual sports will be banned?
I'm quite depressed by this autonomous car business. I enjoy driving and riding, and I see a boring future where we don't have cars anymore. Just very very small buses for 4 passengers. I don't think I'll travel much, since the driving or riding is the most enjoyable part of the trip.
I just hope that the hard part of autonomous driving (you know, the last 20% or the requirements that takes 80% of the efforts) will be almost unsurmountable, so it will postpone it's arrival by a couple of decades (or until I'm too old to drive).
Try it! Library of Babel
Well, I agree that elevated track should be a goal, but it's also scary. When a road is poorly maintained, you steer around some holes. When a track is poorly maintained...
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
But you're assuming the AI driving the car wouldn't also be able to notice an approaching object and be prepared to take preventative action.
The incremental approach criticized here would be letting the human take over when that happens. If the car is taking preventative action, then it's going beyond that.
Suppose you're driving on one of those roads, and find that an accident has blocked the road in front of you, leaving just enough room for you to stop. Do you, as a human driver, then have any options other than preparing for impact? This situation looks to me like one where there is no satisfactory answer, and therefore not a reason to fault automatic driving.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
This is a bad idea even when completely disregarding responsibility.
The "let the driver take over" concept means that the driver will immediately be required to handle an emergency situation without time for proper orientation. If you want the driver to have situational awareness, the driver pretty much has to be driving, because people are simply not good at monitoring situations where they're not doing anything. If the computer can foresee a situation and give a warning thirty seconds in advance, that would probably work (except for the part where a situation that the machine can't handle is very likely to be one the human can't either). It can't in general, because a dangerous situation can come up real fast, so that only a driver who's engaged with the vehicle and keeping eyes out can prevent an accident.
So, realistically, if the computer gives up the car is almost certainly going to crash or do something else bad. At that point, you can assess responsibility if you like, but I'd rather be intact in a drivable car than injured in a smashed one, no matter how the lawsuits will turn out.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
How about the car turning control over to me when I start controlling it? In an emergency, I'd rather not have to perform an additional action, particularly one that requires one of my hands to go to a particular place that's not the steering wheel (or the gear shift, with a manual transmission).
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Well planes are "elevated" and don't have any track at all, but we don't have any problem jumping in those on a routine basis. At least SkyTran tracks are only 20-30 feet or so above the ground, not 30,000, and it's not likely the track is going to fall down. At worse, if there's a problem, you'd have to sit there for a while until maintenance crews come get you; that's no worse than present-day trains (you're not normally allowed to just get out of an Amtrak if it stops on the tracks due to a maintenance problem).
This is a particular problem with small, non-commercial planes. Modern airframes are very safe, and the cats vast majority of crashes are "controlled flight into terrain".
Have you got cites for that? A very brief search turned up this article, which states:
"The vast majority of general aviation accidents in 2011 happened because the pilot lost control in-flight. Another common cause was “controlled flight into terrain,” which means the pilot didn’t see the ground, a mountain, a body of water or another obstacle until it was too late."
On the other hand:
- planes have a few minutes as you suggest, because most of the time the problem isn't "plane is about to explode violently in 2 sec".
Now look at car. The situation requiring a human intervention the most quickly is avoiding an impact. Cars, currently on the road, are already good at avoiding collision. (e.g.: Volvo brand was mentioned in TFA. They have forward looking cameras and lasers. They are good at spoting obstacles and slowing down and stopping before them).
Even with current tech, a car would suddenly require attention to avoid a sudden violent crash. It's already able to slow down safely to a halt. Driver's attention would be required to deal with the unforeseen situation, not to deal with the braking.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
No. My German Aunt {...} Stops in the middle of a left turn
And because the thing happened in Germany, the driver behind managed to brake and stop in time, because they kept enough distance.
In the US she would have been rear-ended.
(in most European jurisdiction YOU ARE REQUIRED to be able to brake and stop. Rear ending a vehicle, no matter the reason, is considered as a ground for not getting insurrance money back).
What the European car maker are working on, is giving a tool to help the *driver behind*. e.g.: A swedish Volvo would have been able to brake and stop on its own.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
Suppose you're driving on one of those roads, and find that an accident has blocked the road in front of you, leaving just enough room for you to stop.
Yep, this just happened to me last week on one of these roads. There was a *huge* accident with a bunch of emergency vehicles blocking the road. I had to slow down rather quickly.
Do you, as a human driver, then have any options other than preparing for impact? This situation looks to me like one where there is no satisfactory answer, and therefore not a reason to fault automatic driving.
Well, there's a few things you can do: if someone looks like they're going to run into you while you're waiting for the accident to clear, you can pull into the other lane, or maybe the ditch if that's a better risk than getting rear-ended. You can also (when no one's coming) try backing up, pulling into a driveway, turning around, and leaving (which is what I did; shortly after I did this a bunch more cars came speeding around the turns towards the accident).
Uh no, drivers licenses are a fairly recent invention in the US.
It's weird, I think, that here in Europe where we have "nanny states" that takes care of us citizens so much, people are expected to be able to handle themselves and be responsible. While in the USA were people expect as little involvement from any kind of government authority expects the pampering and as little personal responsibility as possible.
/ The Arrow
"How lovely you are. So lovely in my straightjacket..." - Nny
Oh please. According to Wikipedia, they had them back in 1899 in NYC and Chicago. That isn't very long after motor vehicles became generally available to the public.
Excellent answer, thanks.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes