Autonomous Vehicles and the Law 417
Hugh Pickens writes "Google's autonomous cars have demonstrated that self-driving vehicles are now largely workable and could greatly limit human error, but questions of legal liability, privacy and insurance regulation have yet to be addressed. Simple questions, like whether the police should have the right to pull over autonomous vehicles, have yet to be answered and legal scholars and government officials warn that society has only begun wrestling with laws required for autonomous vehicles. The big question remains legal liability for the designers and manufacturers as some point out that liability exemptions have been mandated for vaccines, which are believed to offer great value for the general health of the population, despite some risks. 'Why would you even put money into developing it?' says Gary E. Marchant, director of the Center for Law, Science and Innovation at the Arizona State University law school. 'I see this as a huge barrier to this technology unless there are some policy ways around it.' Congress could consider creating a comprehensive regulatory regime to govern the use of these technologies say researchers at the Rand Corporation adding that while federal preemption has important disadvantages, it might speed the development and utilization of these technologies (PDF) and should be considered, if accompanied by a comprehensive federal regulatory regime. 'This may minimize the number of inconsistent legal regimes that manufacturers face and simplify and speed the introduction of these technologies.'"
I Guarantee (Score:3, Insightful)
Re:"Largely Workable" (Score:5, Insightful)
To err is human, to tear down a sidewalk at 55 miles per hour takes a computer.
Or alcohol.
Re:Why wouldn't police be able to? (Score:5, Insightful)
Good news is, since the vehicle is computer based, to pull the vehicle over the police would most likely have to issue a computer command, which could be logged, including date, time and identity of the police officer who issue the other. If it is related to a warrant, it could even be linked to court data.
Re:Why wouldn't police be able to? (Score:4, Insightful)
Sure. But it's still an interesting question. It's illegal for a driver to speed or jump a red light or whatever, but if an automated car with 4 people in it does one of those things, who, if anyone, has broken the law?
Re:"Largely Workable" (Score:5, Insightful)
Re:I Guarantee (Score:3, Insightful)
Re:Why wouldn't police be able to? (Score:2, Insightful)
The designer of the car broke the law, the vehicle is defective breaking traffic laws and needs to be impounded and the builder fined for endangering the public.
When a computer is a box sitting on someone's desk that computes figures and shows lights on a display there is no reason to restrict who can do what with machines and they should be open to hacking and modification. When they are connected to networks the burden goes up a bit and maybe code has to be signed or restricted to a safe API on top of a trusted locked OS (but probably not, in my opinion). But by the time the computer is connected to hardware fully capable of killing people both inside and outside the computer the game has changed and the system needs to be locked down so it can't be hacked and the developers need to take responsibility for their actions. An owner of a car no longer has the right to hack the device because they own it, at least they can't then put it on public roads. Just as drivers need to pass a test the design of an autonomous vehicle needs to pass a test (regulated) to use our roads. This will probably mean leased vehicles owned by the builder company with per mile, per minute, per month fee structures to generate revenue to offset settlements for accidents (which will still happen). The law should then limit the costs of a computer caused accident to the same penalties that a human driver would face for an unintentional accident with the same circumstances.
Re:I Guarantee (Score:0, Insightful)
They'll be crooks, sure enough--but they'll be in uniforms, and carrying badges. The government will even mandate that these cars be hackable. It's for your safety, you know. Then, of course, they'll extend that to vehicles with drivers too...
That's not the worst though. MADD will be out there advocating that you can be prosecuted or drunk driving in a driverless vehicle just because you pushed the start button.
Re:Likely to be adopted elsehwere, far before in U (Score:5, Insightful)
I can see why in the US there is such resistance to autonomous vehicles: Small towns and counties depend on driver error, be it speeding, red light cameras, or stuff like that for revenue. An autonomous system means that everyone will be going the speed limit, so no tickets (and no chance at finding marijuana and thus earning a civil forfeiture prize) will be given.
This is sad because the US is the perfect place for autonomous vehicles -- most cities are too sprawled out for even buses to be reliable, much less light rail. So, vehicles that drive themselves would be ideal because it would allow long distances to be covered with vehicles packed in as much as their computer and mechanical systems would allow, compared to current driving conditions which depend on the driver's ability/reactions (or lack of when compared to a computer.) Even for people who don't own a car, it wouldn't be hard to have a Car2Go/Zipcar like service.
Even more ironic, with computer controlled cars, it would lesson the need for more and more highway improvements. Cars can be sped up or slowed down to allow vehicles in and out, they can be moved into lanes depending on their destination, and if there is a vehicle problem, it can be moved to the side of the road and traffic routed around it without putting the highway out of commission for hours on end. This would save a municipal area far more money than they ever would earn by speeding tickets.
Re:Why wouldn't police be able to? (Score:4, Insightful)
If a vehicle is being operated recklessly, it should get pulled over. If there are outstanding tickets / warrants for its owner, it should be searched / impounded. I don't see why the presence of a driver should matter here.
Re:Why wouldn't police be able to? (Score:5, Insightful)
How?
Is there a database of traffic laws? Who provides the data? Is the data correct?
Does the vehicle read road signs? Are the signs correct? Are they transiently obscured by a parked vehicle or a pedestrian?
Computers, even with perfect design and implementation, are still able to do the wrong thing. Garbage in, garbage out. (I can't fucking believe I have to write this on /. of all places.)
Re:Why wouldn't police be able to? (Score:4, Insightful)
If all else fails, why can't the occupant push the "Computer, please stop at the earliest safe location" button?
(Such a function will be present, as it will fit right in next to the array of "I have to piss/puke/shit IMMEDIATELY" button(s).)
Re:Why wouldn't police be able to? (Score:5, Insightful)
Yes. Every state, online. For smaller locales, the autonomous folk wanting my money had better do a good job of acquiring it, just like the local humans must.
"Does the vehicle read road signs?"
An autonomous vehicle had *better* read road signs, and pretty damned well.
"Are the signs correct? Are they transiently obscured by a parked vehicle or a pedestrian?"
Same problems humans face, too bad.
"Computers, even with perfect design and implementation, are still able to do the wrong thing. Garbage in, garbage out."
Same for the humans, yet fines stand for them. I disagree with your premise. I believe that if a vehicle cannot do all the things a human is required to do, it cannot be an autonomous vehicle. It's just remote-controlled.
Re:I Guarantee (Score:3, Insightful)
No, you can't be prosecuted for riding in a vehicle while drunk. There is no additional risk involved. We ain't gonna see MADP any time soon.
Re:Central Planning does NOT work. (Score:4, Insightful)
Yes Central Planning is bollucks BOLLUCKS I SAY!
As is Central Design for automated cars. Why evolution created an eye so we should just sit on our collected asses for 4 million years and I'm sure an automated car shall simply evolve! And I'm sure as it evolves it'll create the *perfect* solution after those 4 million years.
After all, we all know eyes are the very best possible imaging devices every created. Those silly telescopes, nightvision goggles and highspeed cameras have nothing on our vision!
Re:"Largely Workable" (Score:5, Insightful)
You have a point, but I imagine it will be some time before a "largely workable" system is permitted to operate a vehicle on public roads.
BMW in Germany already have test autonomous vehicles running on public roads amongst ordinary traffic.
But for general use I think it'll happen gradually.There are already publicly available systems that will apply the brakes for you if you are going to collide with the vehicle in front. And systems that will stop you from veering out of lane on a highways. There are even cars already out there that will perform parallel parking for you.
Aircraft autopilots didn't start doing landings from day one. They evolved from much simpler systems. Each step proving itself for a long time.
Re:Why wouldn't police be able to? (Score:4, Insightful)
And what would be the point of pulling it over?
Child stuck in car.
Passenger needs medical assistance.
Bomb needs defusing.
Bridge out ahead, sensors not adequate.
Man, I wish I lived in the perfect world you do. It must be nice where nothing goes wrong, and nobody has any ill intent.
Re:I Guarantee (Score:5, Insightful)
Re:Why wouldn't police be able to? (Score:5, Insightful)
The question goes to the heart of the argument. If the average driver has a 1.5% chance of causing a collision, but an automated on 1% then clearly the automated vehicle is preferable (to over simplify somewhat). However if the 'designers' at GM are responsible for 20 million cars then they have no incentive to ever try and work, because merely by law of averages they're going to get screwed selling millions of cars a year.
A couple of months ago the brakes failed on my car and I narrowly avoided hitting two people. Now the thing is, my car had been at the shop to get the brakes checked and repaired about 3 weeks before that. Who is really at fault? In 3 weeks the auto shop can't really be liable for anything that happened to the brakes, but I had no indication there was a problem until I had a loud thunking sound, and no braking action (go go emergency brakes). Had I been a fraction of a second slower realizing what just happened, well, the law would have held me liable for hitting two people. Even though I would attempt to argue that I did due diligence on the brakes, and was braking from a safe distance (but when you're going 60 Km/h and your brakes fail it takes a moment to process what happened and what your solutions are,and what your fall back scenarios are going to be if the emergency brake doesn't work, and even then you're guessing just how quickly the emergency brake will stop you).
In your case, you're saying what we all know. All data is dirty, and no one thing is 100% tolerant of all possible input cases from the dirty data (in addition to all other failures that can happen on a device). Our legal systems don't really play nice with the real world statistical probabilities of random failures, or how you ascribe blame to something that isn't intentional. It would be most unfortunate if a data entry clerk from 20 years ago is held liable because they typed a speed limit into a database as 80kph rather than the intended 60.
I suppose in some ways it is similar to a national healthcare and medical malpractice problem. People die, all of us. Just as mechanical devices will eventually fail. If you individually mandate responsibility to service providers (drivers, mechanics, doctors) you end up with a much different system than if you collectivize the risk (think NHS in the UK). If the goal is a system that in general reduces accidents you need to move away from trying to assign blame on a case by case basis, and providers who consistently make mistakes can be dealt with internally- but you'll have to accept some sort of shared insurance system for the fact that accidents will happen. Whether that's manufacturers or operators who pay into it (or the government or points of sale or....) I don't know.
The most important problem. (Score:5, Insightful)
The most important problem from my point of view is that the traffic laws don't always actually make sense. Near my house, there's a T junction about 20 meters from a red light. At the intersection, there's a stop line. They used to have a "do not block intersection" sign back at the T junction so that traffic could still turn down the side street. They've replaced it with a "stop here on red" sign. The intent seems to be to create two stop lines for the same red light. The actual law is largely ambiguous on this. It definitely doesn't address this particular situation, but it doesn't say that the town, _can't_ do it. Most people just completely ignore the sign. I just do what I always did before and I don't block the intersection when there's a red light. No-one seems to be able to tell if they're actually required to stop there and, if they do stop, if they have to stop like they'd stop at a red light, or it they have to stop like they'd stop at a stop sign, also, since the turn there is a right turn, and a right turn on red is allowed at a stop light in my state, is a right turn on red allowed there since it's not actually at the light?
So, the problem is the law. It's not logically and consistently written like computer code, it's always open to interpretation. There are many situations where you legitimately can't tell you've broken the law until you've gone before a judge and they've decided. And then, there are many situations while driving where you either have to technically break the law or stop traffic for hours. Consider a left turn at a light where there isn't a separate left turn signal. If the traffic coming in the other direction is continuous, they have the right of way and you can't turn left unless you move to the middle of the intersection, wait for the light to change, then turn. This is illegal. It's what everyone does in that situation and, 9 times out of 10, a police officer watching you do this won't even care. But, consider the situation from a legal point of view. If it's a turning lane, you can't legally change lanes at the intersection to go straight. You can't legally turn left until there's an opening in traffic, which could literally be hours in some places and times, but you can't legally just sit there either, because that's blocking traffic. Aside from that one, there's the fact that you're legally required to stay in a lane unless you're changing lanes, but I've been on a lot of multi-lane roads where the lanes haven't been marked, either because they were faded completely, or because they'd been removed for repainting (months before the repainting in some cases). Legally speaking, all the cars should be grouping into one lane in the dead center of the twenty meter wide stretch of road. That's insane. What everyone actually does is illegally estimate where the lanes should be and travel in them side by side. Then there's yellow and red lights. There are intersections where you cannot avoid running a red light. For starters, you don't know how long the green light and yellow light will last before the red. The guidelines for most states for the length of the lights don't even seem to take the speed limit and the width of the intersection into account and the guidelines often aren't followed anyway. Which means that there are many intersections where, even if the light changes to yellow _after_ you've crossed the stop line, you can't make it all the way across before the red light unless you're speeding. Also, where the intersection actually ends and you're no longer bound by the light is poorly defined both in law and in physical reality. Most people consider themselves clear when they can no longer see the light, but obviously that's at a different point depending on where the light is mounted. Stop lines are another issue. You have to stop at the stop line, but the stop line isn't always in the right place for you to actually see if there are cars coming. Often, you have to stop at the stop line, then move forward (sometimes quite a large distance), then stop again or do a ro
Re:How About Hot-Rodders? (Score:3, Insightful)
There will be an immediate and HUGE problem of folks modding their cars to allow manual override. That should be fun.
Actually, they'll probably all have the option to manually drive them straight out of the box. Think rural environments, dirt roads, navigating your way around a shipping container facility where no map in the world can be up to date enough to help the autonomous car. Also, no one will trust the first models enough to accept a car without the option.
Re:Central Planning does NOT work. (Score:4, Insightful)
What makes you think this will never be changed? Selecting laws here is no different from building a better telescope by trial and error. Somebody has to take the first step.
I think you're making an artificial distinction.
Re:I Guarantee (Score:4, Insightful)
99% of the drivers with satnav at the moment don't bother learning their route, they just follow the instructions that the voice gives them.
My point: human drivers are vulnerable to the same attack vector. I don't expect the research has been done, but I'd bet $1 that if you managed to crack the satnav traffic alert system and fed in data that directed traffic to your dark alleyway, at least 50% of satnav drivers would follow those directions instead of their own idea of the route.
Re:30,000 people die a year in traffic accidents (Score:4, Insightful)
Maybe we can get an autonomous congress at the same time, taking the worst part of government out of the equation - the politician.
Re:I Guarantee (Score:4, Insightful)
How, exactly, would the police pull over an autonomous vehicle if there was no way to remotely access it? Therefore, hedwards was correct: there will be a way to crack the security and force the car to pull over, thus rendering autonomous vehicles vulnerable to the highwaymen. Hmmm...sounds like it could be the plot to a cool sci-fi story...
At this time, OnStar can and does disable vehicles. They have commercials bragging about stolen vehicle recovery. Apparently it disables the throttle. After which the perp presumably guides it off to the side of the road. On th esurface, it seems like a cool thing, though I fear that there will be more and more calls to disable vehicles for more and more things. Deadbeat dad? You come out one day, and the car won't stop.
So it is a trivial matter to expand that to an autonomous vehicle. You will be able to be stopped for whatever they want to stop you for.
But this whole concept begs the question. If we are going to be sitting in vehicles that are guided by something else, if we are just going to be passengers, this is just about as inefficient a way as possible to achieve that little utopia. The line of cars is just approximating a bus. The only advantage is that you have a "last mile" effect of delivering you right to your driveway.
All in all, it seems a little like the olde tyme dream of everyone having a flying car or gyrocopter.
Re:I Guarantee (Score:3, Insightful)
Think "armor piercing bullets".
"Surprise" potholes.
Both the good guys and the bad guys have them.
If the autonomous vehicle could be "hijacked", the bad guy could use the autonomous vehicle to do his dirty work, protecting his anonymity, just as they use hijacked computers on the net.
Think deliberately ramming an armored bank transport with someone else's autonomous vehicle. The bad guy makes off with the cash leaving the bank and autonomous vehicle owner wagging pens and lawsuits at each other.
Just playing devil's advocate here. We do not live in a world where everyone is nice. Anything we make can be screwed up. Its gotta be that way. It guarantees the trodden-under class ways to straighten things out.
Re:I Guarantee (Score:5, Insightful)
The line of cars is just approximating a bus. The only advantage is that you have a "last mile" effect of delivering you right to your driveway.
Except that unlike a bus, it's not occupied by anyone who's not with me (e.g., if they're drunk, I probably am too and don't mind), I can leave my possessions in it, it travels on my schedule (it goes just as many places at 3 AM as at noon), and, of course, it takes me directly to my destination.
Imagine if a 16-hour drive could be done as an overnight trip - you get off work on Friday for a week's vacation, you go home and put the suitcases in the car, eat a bit of dinner, and hit the road. You can be well rested and 1000 miles away by lunchtime the next day. All of a sudden, New Orleans is a weekend trip from DC.
Re:Likely to be adopted elsehwere, far before in U (Score:4, Insightful)
When I traveled around Europe on trains I was thrilled how carefree I could be about intercity travel and how fast and comfortable TGV/ICE can be. Then return to the US and arrive at the decision it is a backward country for dismantling most of its once far-reaching rail network in favor of a car (or two) for every adult - but that's how you get around, which means long trips are a major drag - you have to focus on the most tedius of activities for hours at a time - driving. Ugh. Autonomous Vehicles could alleviate some of this tedium.
I have several observations to make here. First, there are a lot of people who admire European trains, but have no idea how those are paid for. Sure, it'd be nice to have a US train paid for by European taxpayers, like how European trains are funded, but it's a wee bit unrealistic. So then the US would be stuck paying for US trains with hapless US taxpayers. That changes the US-oriented cost/benefit for such projects.
Second, I find it terribly reprehensible to treat infrastructure projects like just another fad. I don't care that you think the US looks backwards for having such an advanced car-based transportation system. It should be, "Does this infrastructure project justify a reasonable estimate of its costs and benefits?" Not, "Uzbekistan has high speed rail so we should too."
Finally, rail projects even in those European countries are notorious for being poor return on investment. And current US projects are laughably bad even by such standards.
For example, it is routine for big high speed rail projects in the US to ignore maintenance and operations costs while grossly inflating ridership estimates. The same politicians who allocate large amounts of funds for construction won't provide for the costs of running that rail, effectively creating huge, long term money sinks for the state and local governments who end up running the system. That's the primary reason that Wisconsin and Florida backed out of high speed rail projects.
Another example, which no doubt will become epic in its extent of failure, is the California High-Speed Rail project. They got a bunch of bond money in the last election cycle and subsequently greatly increased the cost estimate for completion of the rail ($36 billion in 2009 dollars to $65 billion in 2010 dollars). That's a "bait-and-switch" and they have yet to break ground. It also builds poorly used segments first so that the money is spent in a grotesquely inefficient way.
At least, autonomous driving uses the primary strength of the US, it's well-developed road infrastructure and it plays well with what's already there. High speed rail is just a slow though comfortable plane. A lot of its advantage could be eliminated simply by putting in efficient security at airports.