Google Cars Drive Themselves, in Traffic
nytimes.com
nytimes.com
In Boulder, where I live, several crosswalks were identified as being particularly dangerous for pedestrians, and pedestrian crossing signs with big flashing lights were installed at these locations over the past 5 years. This spring, a report was released[1] that showed accidents at many of these crosswalks had actually increased since the lights were installed. "Taken together, the data suggests that approximately eight additional crossing accidents per year occur at these locations," says the report.
There are lots of theories about why, but I think it boils down to one thing: when pedestrians can hit a button and light up these big signs which are supposed to make everyone stop, it makes them feel much safer, to the point that many people will hit the button and start walking almost immediately, without taking the time to make sure that all lanes of traffic are aware of their presence and stopping.
The prospect of automated cars scares me because, obviously, they cannot be perfect, and they will not be able to identify every dangerous driving scenario. Of course, there is a manual override, but I fear that the car being right 99% of the time will lead to such a complacency in "drivers" that, in the 1% of cases where the car is wrong and about to hit something, we will not be able to stop in time. The more accurate the car is, the more safe we feel, and the less likely we are to monitor it as closely and notice when it is wrong.
Anyway, I rather hope I'm wrong. I'd really like to drink my coffee and read during my commute as my car drives me to work.
[1] http://www.dailycamera.com/boulder-county-news/ci_14859190
Two things:
1) neither are people; far from it. (you already knew this though, so...)
2) I wholly agree... as long as other people are driving. Get rid of people's irrationality / drunken behavior on the road, and automated car driving becomes a much easier problem. A computer can take more into account than a human, and decide faster.
The transition is almost guaranteed to occur at some point, unless we reach mass transit for everyone prior to that (and who drives those?). It has to start somewhere, and that start will probably be harder than at any other point along the road.
Even if you have near perfect data about where the other cars are there's always uncertainty about other objects on the road: tree branches, deer, a child running after her puppy. I'd speculate that it's better to rely on a vision system that can recognize those as well as cars well enough to be safe instead of treating different classes of objects differently.
A tree falling on a car while it's driving: relatively unavoidable. Cars hitting that car: far less likely with computers. They don't get "shocked", or hit the wrong pedal, and have faster reaction times. And pile-ups due to an accident are often far more deadly than the accident itself, especially in busy areas where automated driving could make its biggest improvement in safety.
Deer / kids / unexpected obstacles: can be detected prior to them even entering the road by cars ahead, and flagged for caution for cars behind. IR cameras will spot living things for you, motion / blockage detection will spot things near / in the road. Once such a thing is flagged, it's reasonable to assume that other cars would leave more room between each other and around the obstacle, and probably slow down. All things which people should do, but only actually do rarely, and never collaboratively except at extreme cost and increased risk over something organized.
Yes they do. You could perhaps argue they do so less.
Unless they're using a random number generator seeded with outside data, all computer operation can be called deterministic and 100% predictable. Because it's 100% repeatable given identical preconditions. People? Not so / not currently doable.
If the wrong operation for a situation were performed, it'd be due to a higher-level decision going wrong, not the command. Meanwhile, people can "hit the brakes" (high-level command) and instead slam the accelerator (command not carried out). And fully believe they hit the brakes, and that the car accelerated uncontrollably.
I could go for "fully computerised traffic would be significantly safer than human traffic", but absolutes about the former being unfailing are just silly.
And of course, there will be the Android app the pedestrian can use as a last resort to avoid beng run over that stops any cars heading for the phone. :-)
If you're suggesting that we know what a computer will do on every single given input, we don't. That's in a class of unsolvable problems, as it requires the halting problem to be solved.
[Added the following section after being down-voted]
A "shock", in the deterministic computer world, could be an incorrect classification resulting in an incorrect action. The results are similar to when a human is shocked and doesn't act correctly.
Is the scenario I've painted limited to neural nets? No, any trained classifier, even any program at all may have "minor" slips like this. (When dealing with an analogue world, there's a ridiculously large number of situations that may exist). But you know what? Humans don't perform perfectly; and a computer that's been shown to perform the correct final* response 100% of the experimental time, and 99.9997% of the real time is competent at the activity, and trustworthy.
Disclaimer: I love machine learning, I love neural networks, and I see great usefulness for them in this world; however, I also believe that the perfect combination of variables could yield unusual results. Is it unlikely? Incredibly. Possible? yes.
* Consider a situation where a child is playing with a ball. The computer may not recognize that anomaly on the side of the road as the child and the ball, or even as something to be avoided. If the computer recognizes the anomaly as something that needs more information; then tells the car to slow down until it can acquire that information, then recognizes it as something that needs to be avoided and does so, successfully, safely, and with sufficient space for the conceivable outcomes that may have resulted with this child; then, while the computer couldn't discern the object at first, it did in the end and adapted to the situation appropriately. The "final response" was correct.
[edit: rather than simply down-vote me, please tell me where I'm wrong so I can learn.]
I work on autonomous driving robots. Teams I have been on in the past won extremely competitive autonomous vehicle competitions. Our robots crashed all the time. ALL THE TIME!
There is nothing deterministic about how autonomous systems work. In fact our algorithms were by definition random number generators! We randomly generated "hypotheses" and then had heuristic tests that figured out which was the most likely trail and then chose it. It worked well, 99% of the time. But that 1% was a killer.
Mind you, since we were winning we were doing better than the other solutions which tried to be more deterministic and use maps or things like that. We were basically stateless which reduces some of the feedback associated with getting messed up in some of your estimations.
The problem is all at the low level, not the high level.
Give a robot a perfect map and it is easy to find a way around safely. The problem comes determining what the hell the camera is seeing (lasers are a lot easier, but not as flexible and more expensive). That is why the bulk of the effort is focused on computer vision not path finding, path finding algorithms are simple.
Sure actuating the pedals is fairly trivial, but to call autonomous systems "deterministic" is scary to someone who knows how fickle they are. I'm a huge advocate of self-driving cars, but we are not there yet, and when we get there we need to do a pile of peer review on the code (which better be open source). You never know what race conditions will blow it all up.
That's essentially the definition of deterministic. If your machine is powerful enough, you can even do it before the real one completes its operations. Anything deterministic is predictable in this manner.
That you can't predict it, or that you don't have a computer powerful enough to predict it has nothing to do with something not being deterministic, nor predictable. Even a massive, PRNG-influenced neural network is deterministic.
------------
On an aside for people who are thinking of quoting the halting problem at me:
The halting problem requires that the detector returns an answer, and it must be true or false and accurate, or the proof through paradox fails. Therefore, it cannot simulate the operation of what it's detecting, because there could be an infinite loop which would cause it to run without end. You also cannot merely abort it after a while, as that would be either a 3rd return state ("unknown") or an inaccurate response.
But in the field, even if it is stationary the pictures are never the same from frame to frame and everything relies on a lot a randomness, and random noise.
Simulators are almost a waste of time because the real challenge is handling that noise, NOT dealing with the information extracted after the noise. Once you have a clear view of the world, everything it is easy, it is getting the eyes to actually understand anything that is hard.
I have made no claims that computers can make flawless decisions. I make no claims they can detect / avoid everything avoidable. I don't know where you're getting the idea that I've said this, unless you're dragging in past battles with other people - everywhere I've claimed computers would do better I've included qualifiers like "likely".
The point I was trying to make is that the instructions of a computer, barring external phenomena which cannot be accounted for, are fully deterministic and accurate. If the high-level decisions are made incorrectly, then the end result will be incorrect, but all down the path everything will be accurately performed.
Not so with humans (just look at someone with epilepsy, or at your own memory), or at least not in the minds of some people, which is the only reason I include it. And, if nothing else, we're far further from accurately simulating an entire human body / neural net than from simulating another computer.
Besides. I never said nor implied "perfect". Merely better, and being better is easier without other human drivers, who can far more quickly take a bad (or good) situation and make it worse. Bikes don't go 60+ mph and weigh over a ton. Neither do pedestrians. And very few of either of those carry large quantities of gasoline.
Not only does the technology have to become better in the aggregate than humans, it has to become so much better there's aren't any crash situations where human drivers perform better. There also can't be many crash situations when the computer doesn't significantly outperform a human. Otherwise both public sentiment against the cars, and legal liability will be too great.
Imagine: whenever there's a software bug and an automated car crashes in any situation where a human might not have, the GoogleCar manufacturer will get sued, the local paper will write an article about how Google killed someone, etc. And then, once a year, there will be another story about how crashes are down 2%... that's a lot of lives saved, but which article do you think will get more attention? And do you think "our cars usually make driving safer, just not the night of October 10th when John was killed" will be a winning defense?
I can't wait for this technology, I think it really will deliver all the benefits the article mentions long-term. But it's a long way off (I hope I'm wrong). I give a lot of credit to Google for investing in it now--this is one of the things I read and think "don't be evil" is more than just marketing-speak.
If there was a automation feature that prevented accidents in 90% of the situations where a human would crash but caused an accident in a situation that was 10% as likely to occur as a human crash, the result would still huge liability since the manufacturer would "definitely" be at fault in the latter case.
Normally, the programmer is in the clear but obviously there would be exceptions - if you could prove they intentionally introduced code that make the car do something bad, they would be liable. Now introduce a plaintiff's lawyer to muddy-up the definition of these terms.
Let me point out that determining liability/culability/etc in a messy accident is difficult task when you've got just two human decision makes involved. It's decided by jurors who aren't necessary geniuses but probably think they understand human reactions and decision. Introduce two or three more decision makers into the discussion and you've got a messy process indeed.
We saw how far the faulty Toyota accelerator pedal charge went. A big factor in the reactions to evidence was the spookiness of "drive by wire" (I think we've gone over how the actual cause was driver fault but consider how the complexity of the system made a lot of people less likely to believe this).
I mean, it's terribly ironic. Driving in general is one of the most dangerous activities a modern person can engage in. If robots drove our cars the way we drive our cars, the protests would be huge. In a sense, the way we've socially normalized driving is through the huge insurance and liability complex. This assures people that when someone drives wrong, they will be punished somehow. A similar or even significantly small number of accidents without person to punish would be unsatisfying socially.
Besides, humans are really lousy drivers. They should not be trusted to operate such vehicles at all. Machines will makes mistakes too, but I bet we'll see a lot less of them compared to the era of human pilots.
You mean speaking, right? ;)
I go food shopping every week at about the same time. I email trip plans between friends and schedule them. I don't drive to work but certainly if I did that's an obviously predictable event.
Between synergy among the various google properties and good pattern-recognition in the google cars (or whatever) system itself I'd be surprised if predicted trips didn't hugely overwhelm explicitly-ordered ones.
They can, at least 1000 times more perfect that humans are. A computer could have and exact model of the road, and compare it to what it sees, could report bugs in an objective way, the same way computers on airplanes work, having made flying way more secure.
Crosswalks are a problem because you can't predict car behaviour: some will stop. Some will completely ignore them (for reasons unknown). If there were no crosswalk you know exactly what cars will do. That's much safer for both parties.
The "illusion of safety" is most dangerous at traffic lights. 2-3 times I've had cars just sail through, completely oblivious. Note: they're sailing through on RED lights. Not amber going red. It's not one of those borderline cases.
Once I had a cop have a chat to me about jaywalking but I've never been fined. I would be pissed if I ever was. In a fight between a car and a pedestrian the pedestrian loses (big time) so pedestrians have a higher vested interest in their own safety. I know what the light changes are at intersections I cross a lot. Where are these cops when cars sail through red lights or nearly run over pedestrians when turning when the pedestrians have right of way?
I get the distinct impression these skills will come in handy when I move to New York next month!
Anyway, back to the self-driving cars: this I believe will be a painful transition that will take many, many years. It's nice to see Google working on this.
Look at how many people drive off cliffs, into trees, or into other cars when a deer runs across the road. Vested interests in safety aside, it's always dangerous to do something unexpected in traffic.
So pedestrians can do whatever they want because its more dangerous for them? Did you know that a flashing red hand means "don't walk"? You can be ticketed by entering the crosswalk when it is flashing. How many of your close calls would that eliminate?
As a driver, what is your first instinct when you see someone in the middle of your path on the road? I also hate people who jaywalk onto medians - how do I know they're going to stop when they get there? Especially if they're running at full speed. I've also had several pedestrians almost jaywalk right out infront of my car before, some not noticing me until I honked at them (too close to stop).
We would all be safer if we all followed the rules. Not just cars.
Many people seem to neglect or completely fail to understand the risk of increased perception of safety leading to an overall decreased safety in these systems. We can't have the best of both worlds – a machine's constant attention and reaction time coupled with a human driver's intuition – because the moment you give him an autopilot that he thinks will keep him out of trouble, the human sits back and stops paying attention to the road.
These automated systems may eventually become all-around better than any human driver, but it will be take many iterations to get there, and the first versions will certainly lack a human's ability to anticipate trouble before it arises. It will be a long while before an autopilot can say "hey, that car ahead is weaving unnervingly, the driver might be drunk so I should keep my distance" or "the cars ahead of me are curving slightly to the left, there might be a cyclist riding on the shoulder so I should slow down a bit and do likewise." And until we have that level of automation, I'm afraid our roads are about to become a lot more dangerous.
What makes me feel unsafe are the drivers who break the law (you don't know who you are) and the drivers who hate cyclists and pass us as closely as possible, throw bottles at us, etc (you do know who you are).
What makes me feel unsafe is the unpredictability. I would expect a computer would just follow the rules exactly and be very easy to bike around.
Anecdotal evidence is this blog, the link is to the first of a growing number of entries. http://www.101wankers.com/?paged=2
There might be a startup in there somewhere.
My gut tells me that drivers in unpopulated areas are less aware and less able to 'share the road'.
I was struck by this comment, because I do this every day...by riding the train. Automated cars, flying cars, double-decker highways, etc. are fantastic innovations, but I sometimes wonder if we couldn't do better than the car, at least in terms of our regular commuting. I ride my bike and take the train because I have that option where I live and I love it.
In a car, you can choose your own route, which makes a lot more routes feasible to do in a reasonable amount of time. Plus, in an automated car, you can spend your entire trip reading and drinking coffee, instead of having the walk to and from the train, trying to juggle your coffee and umbrella and laptop all while trying to hurry to catch the train.
It seems to me, at some point, computers will easily out drive humans.
One thing I read said the human driver had to take control because of a cyclist. Obviously there's more work to do.
But eventually these systems will get there. And they'll probably be able to share information with other vehicles too. So robotic vehicles will be able to see around corners. And tell each other exactly what they'll do before they start doing it.
Safety could wind up being 6 sigma instead whatever is currently. And we'll actually have the ability to measure it because collecting such statistics will be more feasible.
Right now, most of our screw ups don't end causing accidents and they never get recorded. And we prefer it that way! But we'll be able to do data mining on every little thing robotic cars do.
Which is not saying this effort shouldn't continue. Human drivers don't always handle cyclists well either.
unfortunately, I think a lot of Americans are already doing this, or something close to it, today -- but without an AI driving the car. I've lost track of the number of times I've seen women driving huge SUV's with kids in the back, and the woman is yapping on a cellphone while checking her makeup in the mirror, etc. Oh and, while driving.
My biggest fear is that there will be a manual override.
So, although I'm hugely supportive of this, and I think it's possible to have fully autonomous vehicles within a decade, I want to see it. In production. Then I'll believe. Up until that point it's like one of those science stories where it works in the lab but nobody has any idea whether or not it will work with humans. There is a long, long way to go from the mechanics of AI and auto-driving and the real thing happening. It's not just the technology. There are about 17 large groups of people and organizations that need to be re-aligned before this would ever fly, even if it were perfect. This could easily end up like pot legalization, where it's obvious for years that the current laws are idiotic but it takes an entire lifetime to get the politicians on the same page as the public.
I'd love to go all blue-sky on this, going on and on about the massive changes true auto-drive would entail. But instead I'm opting to be very cautious about separating press releases from real products and benefits.
(Still, it's awesomely cool)
I think one of the key issues for Google, from a corporate strategy standpoint, is "freeing up people's time." Driving is one of last places where we spend significant time awake without being able to use the Internet and hence any of Google's services (except if you use a smartphone, which is now illegal in some states, and in any event isn't an ideal place to be clicking on ads). There is a safety issue here as well. Although using a smartphone while driving is illegal in some states, people are driving while using their smartphones with increasing frequency. We need our Internet "fix."
I am sure this driving technology also taps into several of Google's key capabilities: e.g., programming expertise, its voice recognition technologies, search, and its mapping software (Google Maps and Navigation).
It sounds like this may be as much an idealistic pet project as it is any kind of corporate strategy.
A company like Google can turn an idealistic pet project into something truly game-changing with consistent time and resources.
Crazy stuff Google did:
* mapped all the world's roads using cameras and a mechanical turk, destroying billions in market cap for map data providers
* Android, groundbreaking in both technology and pushing the envelope on patent law
* launched a satellite
* deep tendrils of fiber optics and community oriented wi-fi projects to connect the world
It doesn't surprise me at all Google is building autonomous cars. Search and adwords are not even the most intriguing things to me about Google these days.
Imagine a mad scientist who has $12,000,000,000. That is Google's strategy... also lots of testing.
Can't beat that.
And even though it's not a core part of their business today, the first company that can offer a viable automated trucking platform is going to be strongly positioned in a multi-trillion dollar industry.
People criticize Google for being "just an ad company". Branching out with tech like this would put an end to that pretty quickly, wouldn't it?
A human-driven Google Street View car would not only retrieve mapping data, photographs for the Street View and cellphone tower and wifi signatures for positioning; it would also serve to annotate the roads for easier navigation by autonomous vehicles.
(personally I think the other comments about google doing it mostly because it is a hard, cool, worthwhile challenge are most on the money though)
The insurance company, just like today. Possibly paid for by the manufacturer and entirely included in the sales price, or possibly paid for by the driver/owner like today.
duh, of course! The question is - at what premium? Because you know, actually it's customer in the end who pays the price.
Also fully automated taxis start to look like a mass transit system, and if Zip car offered to pick you up at your house...
Of course, there also will be a government required back door to stop cars that are being used in terrorist attacks.
IANAL however.
Suppose there's an extreme edge case where a bicyclists fluorescent jacket appears as a road to the car; so the car chases down the cyclist and runs him over. This is not a hardware failure, but an active decision made by the car to cause injury to a person. So in such a situation, who would be responsible? I guess it is a question for technolawyers to decide.
I'm not saying there are no details to work out. I'm just saying that it's not some unfathomable chasm we have to cross. We deal with similar questions all the time right now.
If the automated system is less error-prone than a human, the driver can't reasonably be negligent for doing nothing ("it would be incredibly foolish and arrogant for me to think I could handle that situation better than a computer designed by 10 Phds, especially when the insurance company agrees that the computer drives better").
For the manufacturer the reasonable position would be that as long as they followed certain verification/testing standards, and fixed any discovered bugs as fast as reasonably possible (probably not issuing a recall or telling people to turn it off, assuming the system was still safer than a human driver), they should be fine. The industry might need to get together and come up with standards and a certification body for the first part to work, and I have no idea how well the no-recall thing would work without something explicitly permitting it.
The prosecution would likely argue that no computer program can be a match for a human's ability handle new and surprising situations: the "driver" must still pay full attention to the road, whether a computer is driving or not. It would not be enough for a computer to be statistically better at driving, it would also have to not make mistakes that a reasonable person wouldn't make.
Also, we don't have information on how error-prone this system actually is. All we know is that a car it was driving had one accident. We don't know how many times the human driver had to take control because it would have otherwise crashed.
[obligatory IANAL]
I suspect that this ability is due to our having more awareness of context. I imagine it wouldn't be terribly hard to dump a larger amount of contextual knowledge into a car, probably just attach an expert system and pay a few tens of millions of dollars to have people fill it with data. I'm not sure what the state of machine learning is, but they could perhaps pay owners to upload logs (maybe especially after any accident / close call, or maybe that would give a skewed perception) and then crunch those to find new rules.
> Also, we don't have information on how error-prone this system actually is. All we know is that a car it was driving had one accident. We don't know how many times the human driver had to take control because it would have otherwise crashed.
Yup, that the humans ever took control at all says it probably isn't quite ready yet. But it looks very promising. If I ever have kids I might not need to deal with them learning to drive, and maybe my next car could let me browse the internet on the way in to socialize (well, why else won't they let me telecommute?) at work.
I don't think that's the case, even taking as a given America's disgusting obsession with lawsuits. There's an obvious answer to the liability question, and one that I think Google is already depending on in allowing their robot cars to drive all over California: The person in the driver's seat is responsible as always. That person can choose to press the autopilot button but it's not fundamentally different than cruise control. Touching the brakes (or the steering wheel) immediately transfers control back to the driver. The car is doing what it does, as always, strictly as the driver's proxy.
Now of course it will happen that some driver sues the manufacturer because there was an accident they believe wouldn't have happened if they'd never gone on autopilot. Just like there have been (I blithely presume) lawsuits over cruise control. But the manufacturer will weather those lawsuits and reasonable legal precedent will be established soon enough.
But now suppose you press autopilot, go to sleep, and wake up to find you've committed vehicular homicide. That's a scary thought, but is it a potential dealbreaker in bringing this to the mass market? I think the answer is no, because we've already established the liability question: You were, officially and legally speaking, criminally negligent to go to sleep. Unofficially it might well have been a perfectly reasonable risk to have taken, if the probability of it ending in tragegy is sufficiently low. Logically, you could argue, citing statistics on human and robot drivers, that going to sleep with the car on autopilot was markedly less negligent than not putting the car on autopilot at all. I don't know if that would fly in court. But as far as legal obstacles to adoption of robot cars, I don't see it as much of an issue.
PS: I'm not a lawyer.
Cancel the bullet train, and instead plow all that money into rapidly deploying this.
Imagine a future where cars are not owned by individuals but are owned by few companies (or maybe even government, the way public transport is) and participate in what is essentially a global taxi network.
Instead of using your own car you would just step out of your home, enter destination address into your phone and press a button. That would be sent to a server which given your location, position and destination of all existing cars in route would pick one that is the closest to you that has available seat and goes in that direction. That car would just pull up and you would be notified in real time how far it is from you.
If a car is not available, it would be dispatched from the closest holding area.
Think ZipCar except with cars that come to you, drive themselves and you can share with other people while it drives itself (or you can have it for yourself, if you prefer, except at a higher cost).
This kind of a network would drastically reduce number of cars needed to drive people around and given global knowledge of where people are and where they want to go, it could plan the absolutely most efficient route.
For popular routes (e.g. if it detects a pattern that a lot of people commute from San Francisco to Mountain View around 9 AM on Mondays through Fridays) it could deploy bigger cars (i.e. buses), further reducing traffic.
When you think about it, it's not surprising that car companies will not work on a technology like that since it would drastically reduce demand for cars and if this comes anywhere close to reality, expect lots of legislative and political battles (after all, this will mean that Americans in Detroit will loose jobs).
What's so beautiful about it is that it takes a massive infrastructural liability -- a century of building our world around the automobile instead of around rail or other mass transit systems -- and turns it on its head. We get out of infrastructure jail free.
Think of the time humans spend behind the wheel of a car today. How many hours a day? How many solid days of driving a year? How many man-years of humanity are wasted on fruitless idling in traffic? A society with autonomous personal transport becomes a measurably more productive one and likely a happier and more satisfied one as well.
Car companies may fight it but it won't matter, the economics are overwhelmingly against them on this, because signing up for this service will be overwhelmingly cheaper than owning your own car, probably by at least an order of magnitude for most people.
Random digression, but I think you will find this is an NP-hard problem - I'm quite sure you can encode the Travelling salesman problem[1] within it. So the absolutely most efficient route is probably not feasible, although the system could certainly do a good job optimizing over what we have now.
[1] http://en.wikipedia.org/wiki/Travelling_salesman_problem
Personally I view mass transport as the stop gap to deal with the fact that cars really suck today. Mass transport will never be optimal because (a) it is mass transport, which means sharing space with a bunch of (potentially sick) strangers and (b) the bigger the group the less common the final location. A car can go exactly from point A to point B but a train can only go between stations. Every change between train (city-to-city) to tram/bus (location within city) is wasted time, as is every tram/bus stop that you don't use.
In my perfect world we would get rid of mass transport and just have a sort of network of cars. Most trips people take are planned well in advance so you could schedule when you want to leave, walk outside and step into the car that's waiting there for you, kick back and wait to be let out at the door of where ever it is you're going. The car wont need to park because when it's done with you it goes for the next person.
Something like this would be much more efficient than mass transport that has to just keep on making the same trips regardless of how many or how few people are using it at those times because people might need to use it.
There was a period before cars where trains were the most efficient form of long distance travel, so cities were built around this fact. Then cars got popular, so we built cities for those. Then they pushed light rail to combat suburban sprawl. It would be funny if automation made cars the ideal mode of travel again.
There is one big problem - convincing some people this is true, and that cars will make mistakes, but far less mistakes than humans.
The problem is there are many people that fail to see it that way. More than once I was in a situation of explaining the concept of automated cars to others, and always the opposing arguments go something like: "But, can you be absolutely certain that there isn't going to be one single case where a car makes a mistake and kills someone". No, I cannot, and I will not.
And, of course, once the self-driving cars become a reality, such cases will happen. And than there will be an article in some scandal-seeking newspaper how an automated car killed the father of two, and how this is a what happens when you let science control your life and when you let all those over-educated people do what they want... Or something along that lines...
Frustrating.
Automated cars eliminate many of the costs of driving--lost time, frustration, etc. When you decrease the cost of an activity, people are going to do it more, not less.
In fact, if you ask people that can afford cars but commute by train why they do it, many will say that it's because they can work on the train and not have to deal with driving in traffic. Those people would definitely consider switching from the train to automated cars.
"Hey, traffic's not as bad as it used to be since the Google auto-drivers smoothed out the traffic. Let's go to the store at 5:00 today!"
Congestion will again increase to the same point of suckage as before, but then, only non-human drivers will be qualified to be on freeways.
I'm in NYC, I could hop on the subway any time I want without any hassle at all, but it's not like I do just because the option's there. Transportation is a means much more than an end for most people.
Uh, maybe you'd have to look at the history of urban planning for the past 50 years? In densely populated urban areas, just about every time people add more roads, the capacity gets used up by increased development in the outlying areas served by those roads.
Network bandwidth and CPU capacity are subject to the same phenomenon. Do end-user desktops really have more functionality than 10 years ago? Besides things like increased 3D graphics capability and more things to do with a web browser and an internet connection, not so much. As bandwidth has increased, the amount of data in a webpage has increased, and it becomes practical to use a web browser for more.
But don't take my word for it:
http://www.assmotax.org/Releases/AMCT%20release:%20building%...
http://www.walkablestreets.com/widen2.htm
http://bicycleuniverse.info/transpo/roadbuilding-futility.ht...
Basically, there can be a certain amount of pent-up "latent demand" which is there but which can't manifest until the roads are built and such travel becomes practical.
But again, it's just a guess, and I'd be the first to accept that I was wrong if the numbers didn't actually work out.
In the PR post two savings mentioned are "highway trains" and car sharing. Car sharing gets interesting if the car can drive by itself to pick up the next person, to leverage all the normal downtime in a typical day where your car is just sitting there waiting for you to use it next.
For those of you on Netflix, it's available as a streaming video [2], and I highly recommend it.
[1] http://www.pbs.org/wgbh/nova/darpa/ [2] http://www.netflix.com/WiMovie/The-Great-Robot-Race-Nova/700...
Google robot > Humans
Anyway, I would expect that this issue will be looked into a bit deeper once companies start producing autonomous cars.
Truly looking forward to the commercialization of technology.
“The technology is ahead of the law in many areas,” said Bernard Lu, senior staff counsel for the California Department of Motor Vehicles. “If you look at the vehicle code, there are dozens of laws pertaining to the driver of a vehicle, and they all presume to have a human being operating the vehicle.”
The Google researchers said they had carefully examined California’s motor vehicle regulations and determined that because a human driver can override any error, the experimental cars are legal. Mr. Lu agreed.
Also, this has great potential (though would probably generate a lot of controversy) for drunk driving. Imagine if your car had a safety mode that could watch out for potential accidents and prevent them even with a human driver.
Amazing possibilities ahead.
Well, if my car would do that I'd certainly drive to the bar more often. I'd want the fully autonomous "drive me home, Jeeves" system though.
The future is now, man :-)
http://jacquesmattheij.com/Autonomous+driving+cars+one+big+s...
The google car seems very much reliant on its gps data, in the article there is this passage: "He did [interrupt auto mode] so twice, once when a bicyclist ran a red light and again when a car in front stopped and began to back into a parking space. But the car seemed likely to have prevented an accident itself."
Seemed likely is not really good enough though, you have to be sure enough not to interrupt the auto-driving at all for this to be usable.
The biggest problem with auto-driving cars that have a human supervisor is that there is nothing more dangerous than having to be alert for a long time when nothing happens. Invariably you will be distracted when something does over the course of tens of thousands of miles.
Regular traffic is one thing on mapped streets is one thing, but that 100's of thousands of emergency situations that a normal career on the road throws at you over a lifetime contains quite a few where I really wonder how a 'robot' driver would deal with them.
I'm all for 'automated drivers', but I'd like the researchers to be sitting in the passenger seat and to have their kids playing around the car (or come out on their tri-cycle from between two parked cars) while they drive it to prove that it is safe. Overrode but the car would have probably done it is not good enough yet, even though the achievement is very impressive this is not as close as it seems.
Computer programs are great at dealing with everything that you can think of beforehand, it's the exceptions and the response to those exceptions that matters in an application like this.
I also wonder how they'd deal with liability issues if their driver was not actually paying attention to traffic and the software made a mistake and caused an accident.
Seems like quite a few people are working on projects such as these. I heard some hairdo on TV the other day musing that they will reach general availability "as early as 2022". How do they even come up with those numbers?
They pull them out of their asses. :)
Don't think I like the sound of that...
>The only trouble with this set up is the car will be incapable of reacting to unexpected obstacles, say a boulder rolling onto the road or a spectator jumping out to get a picture. Careful spectators.
Seems they agree.
Driver, there's items you like on sale at that store. Driver, I think we should stop and check it out. Driver, we are stopping.
Dear, I don't feel safe having Lisa drive with that new Mark boy. You can just tell he dials aggressive.
But seriously, from a marketing perspective, I think the low-hanging fruit here is highway driving. It's a small, logical and therefore adoptable step up from cruise control, just with a little auto-steering (they already have auto-braking IIRC), and perhaps some networking for predicting lights, congestion and co-ordinating with other (online) cars.
Start with an "online-only" lane.
So much of the groundwork for what Google is doing today was already in place three years ago thanks to DARPA. Google's contribution, as far as I can tell, has been to pour more funding into the project and contribute its extensive mapping data and its phenomenal capabilities in collecting and crunching huge datasets.
When fully autonomous road transport does finally become a reality, it'll owe as much for its inception to DARPA, with its innovative X-Prize-style Grand Challenges, as it does to Google. And naturally to all the competing teams who over the course of three DARPA challenges drastically advanced the state of the art and overturned previous assumptions about what was believed possible.
This is a fascinating example of academic research, innovative government funding and long-sighted private funding combining to create something really cool and useful. I hope this sort of thing will continue.
Police will be able to force anyone to pull over anywhere, at any time, and for any or no reason.
Even though they can do that now, people still decide to pull over. They are not forced to.
And what if there becomes a database that police forces compile, showing the history of movements of everyone? That database could potentially be leaked.
I think the privacy implications of this should get at least some thought.
Of course, with computer-controlled cars, you can eliminate traffic infractions and hence most if any need for police to pull you over anyway.
Or they'll develop their own and sell it through a third party.
Have they tested it in the snow?
Have they tested it driving on ice?
Does it know when its fuel gets low?
Can it tell when a head gasket blows and the engine is overheating?
Have they driven it on the highway?
Have they driven it on the countryside?
How does it treat cyclists?
How does it treat pedestrians?
How does it treat deer, squirrels, birds and other fauna on the road?
How does it respond to aggressive drivers?
Can it park?
Can it parallel park?
Does it know not to block the box?
Does it know what to do when a traffic light is out?
Does it know to pull over when emergency vehicles need to pass by?
Does it know how to merge?
Does it know how to avoid hanging out in other cars' blind spots?
Does it know that eighteen-wheelers have large blind spots ("If you can't see me, I can't see you")?
Does it know what to do when four cars come to a four-way stop intersection simultaneously? (Does anyone?)
Can it find parking?
Can it pick a parking spot that's far enough away from driveways and fire hydrants?
How much of the logic is dependent on the car it's installed for?
How much work is involved in installing it to a new car?
What does it do if it gets boxed in by FBI agents on the highway?
What does it do if debris knocks out the camera?
What does it do the if car in front of it starts fish-tailing on ice?
Sorry, how do YOU know when your head gasket has blown and the car is overheating? What's that, your first indicator is the electronic temperature sensor on your dashboard!? My, there certainly is no way a computer could detect that.
If you are about to say, you detect it through loss of power, steam from the engine bay, coolant in the oil... I got news for you, you just detected it slower than a computer would have. A properly designed automatic driving system would monitor the vitals of the engine, and pull over in the event something goes critical, much the same way your PC shuts off when the CPU hits some predetermined temperature. Even better, if all cars were automatically driven, it would communicate this abnormality with the other cars on the road. It would be able to determine that this is unusual (it's not just a hot day), and it would also be able to co-operate with the other cars on the road to pull over safely as fast as possible.
The things you list are not "this is why automatic driving is impossible". They are just "here's special cases we have to make sure to catch". In the case engineers miss a case, good feedback control systems can often cope for the unexpected event, and in the event they can't, there's always manual control. Hell, let's be honest here- most drivers would not detect a blown head gasket until the car simply stopped functioning, and the engine was completely destroyed anyway. There's tons of case where humans perform terribly, including some on your list. Accidents are caused all the time by people braking for squirrels when they shouldn't.
You are exactly correct; this is actually all I meant to express in my post. I suppose I should not be surprised that my comment was taken to be espousing one of the prevailing extremes! Sorry about that--I should have been a little less terse.
> most drivers would not detect a blown head gasket until the car simply stopped functioning, and the engine was completely destroyed anyway.
Ironically, this exact thing happened to me at the end of last year. That's actually why I included it in my list... not because I think it's un-doable, but because I really really hope it's an edge case they cover! :-)
Could Google shorten the timescale for this massive shift to just a decade or two instead of most of a century?
http://www.youtube.com/watch?v=aNi17YLnZpg (volvo crash avoidance failure)
to shake people's confidence in these systems.
what about if GM announced a search engine? Everyone would have their opinion then.