By on January 23, 2012

Google’s nutty pseudo-utopian autonomous car project faced a reality check at a legal symposium sponsored by the Law Review and High Tech Law Institute at Santa Clara University. Among the challenges raised were the prospect of insuring such a car, and whether the car would be able to stop for law enforcement or construction workers.

While Google claims that their autonomous cars have driven more than 200,000 miles  of accident-free driving, issues like whether police can pull over autonomous cars, as well as technological limitations with artificial intelligence, still remain as stumbling blocks. Google is throwing a lot of time and energy into having laws changed so that autonomous vehicles are road legal, but based on the concerns raised by experts, it looks like self-driving vehicles still have a long way to go before becoming viable.

Get the latest TTAC e-Newsletter!

53 Comments on “Google’s Autonomous Cars Face Legal, Practical Challenges...”


  • avatar
    Andy D

    The first automobiles in England had to be proceeded by a flag man. As long as the car has a good insurance policy, what is the trouble?

  • avatar
    tced2

    When the Google autonomous automobile makes a mistake – driving over the speed limit – or worse, an accident killing someone – who pays? Does Google pay for the coverage? Despite the Google bank account being very large, I seriously doubt they would take liability for every Google autonomous car. the “200,000 miles driven” is a drop in the bucket of miles driven by all automobiles in the US. I suspect the “200,000 miles driven” is not statistically significant enough to declare the Google autonomous car a good device. It needs much more testing.

    • 0 avatar
      nonce

      200,000 miles without an accident is definitely better than the average human driver.

      Obeying the speed limit is what a google car would be perfect for. I might do so too well, so I wonder if it decides to keep up with the flow of traffic if that’s over the speed limit. I know that it had to learn how to be aggressive enough at a 4-way stop in order to get its turn.

    • 0 avatar

      The insurance company will pay, just like today.

      Sure the first few hundred or thousand vehicles may need to be insured by Google themselves, but after they have a few hundred million miles logged the insurers will have actuaries look at the claim data for Google driven cars and set the rate, just like they do for 37 year old, married, male, 2 infractions, 0 duis.

  • avatar
    Landcrusher

    200k is crap.

    First step is selling this as an autopilot which the driver is still responsible for at all times. After 100 million miles with average safety should they try to fix any laws.

  • avatar
    econobiker

    “Google is throwing a lot of time and energy into having laws changed so that autonomous vehicles are road legal”

    Remember folks, the money is not in autonomous automobiles (maybe some money in it for the insurance companies telling you to program your car to drive their way and only their way) but the real big money is in autonomous tractor trailer trucks. I am thinking that Google is using the cars as proving grounds for the truck industry. Companies that eliminate truck drivers from their logistics system stand to save big bucks both in salaries and liability.

    Wal-Mart, for instance, would probably convert to autonomous trucks as soon as that firm could legally do it.

    • 0 avatar

      True, trucks could be running 24/7. No sleep regulations, no paying drivers.

      But I do think there is money with automobiles. People wouldn’t own autonomous cars. They would sign up for a service that guaranteed cars on demand (hailed by smart phones.) No maintenance, no insurance, no problem getting a ride home if you’re drunk.

      It would be a monthly fee that guaranteed you a car when and where you needed it.

      • 0 avatar
        Sinistermisterman

        Holy crap! I think they have something similar to that in existence already… isn’t it called… a Taxi?
        I can see your point, but even so I think that many people would rather have the hassle and expense of their own car rather than ride in a cab everywhere, even if it was autonomous.

      • 0 avatar
        nonce

        Taxis aren’t economical because you need to pay for the meatbag that sits behind the wheel.

        On the other hand, my neighborhood of 60 homes could have a private fleet of, say, 10 cars. They would still drive right up in my driveway to pick me up.

  • avatar
    Sinistermisterman

    Although a long way off, I’m pretty sure that this technology will be the future for 90% of driving. By getting this technology in the bag now and pushing for legislation change, Google will be sitting pretty when they get to license this technology to manufacturers.

  • avatar
    Russycle

    I don’t see why people are brushing off the 200K incident-free miles. Supposing your average driver does about 10K a year, that’s equivalent 20 years accident-free. Also known as an insurance company’s wet dream.

    I’m not saying a lot more testing isn’t needed, but it’s a hell of a good start.

    • 0 avatar
      imag

      It’s amazing how quickly we also get jaded by the incredible. I mean, a fucking computer drove on our roads for 200K miles without an accident. It’s incredible.

      And if it does get into an accident, we can actually fix the problem and prevent it from happening.

  • avatar
    nonce

    You don’t want the first generation to be autonomous. You want the first generation to still have the meatbag (aka “driver”) behind the wheel, just because it’s easier to deal with from a legal perspective.

    Like SrMrMan, I think this is going to be the way of all driving. Possibly within 10 years. The change will happen really slowly, and then really fast, once people realize that on net computers are better than humans.

    I’d pay a few thousand right now to convert my car to autonomous.

    • 0 avatar
      Zammy

      I disagree. I think it is easier to deal with from a legal perspective if there is no human in the vehicle.

      As long as their is a driver/operator in the vehicle, law enforcement and the judiciary will be faced with having to decide to what extent the driver/operator is responsible for accidents and/or traffic violations.

      If you completely remove the driver/operator from the vehicle than the decision becomes much simpler as their is simply no person physically present to hold legally responsible.

      Once the legal system works out how to deal with this, then you can insert the driver/operator back into the system but it is clear that they are basically cargo. As long as the person isn’t interfering with the autonomous operation of the vehicle there is no reason that their presence would change anything.

      Now I’m curious as to which states have laws that even require a driver/operator to be present. It is one thing that have a bunch of laws that *assume* there is a responsible party present, but I wonder which states if any *require* there be a responsible party present.

  • avatar
    jmo

    The key is the system doesn’t have to be perfect – it has to be better than the median driver. Not all that high a hurdle – IMHO.

  • avatar
    mitchw

    The Horror. The Horror. If Google autonomous cars drive around at the speed limit, think what this means for someone driving behind said dustbusters. Who goes 25 in a 25mph zone without going bonkers? We need some dirtbag lawyers to make Larry and Sergei’s silicon dystopia go away before it ever happens. NERDS!!!!!

    • 0 avatar
      protomech

      Given the choice between driving at the speed limit or driving safely (at road speed), the autonomous agent will probably choose to drive safely. And not in the left lane, except when passing.

  • avatar
    "scarey"

    You are being detained under the National Defense Authorization Act of 2012 and the Proactive Crime Prevention Act of 2013. “Step into the van, or there will be…TROUBLE”.

  • avatar
    laphoneuser

    I can’t wait for this technology to be in FULL force on the Los Angeles freeways.

  • avatar
    KixStart

    I’m looking forward to it. Every year, I take a few 500 mile trips and a 3000 mile trip. That’s days of my time wasted. There’s no driving fun in any of that. If I could turn piloting the vehicle over to an autopilot, I’d love it.

    It wouldn’t even have to do the whole trip… I’ll get it onto the expressway and retake control at the exit.

    Heck, it would even shorten the trip by a day and save me two nights’ hotel bills. I don’t drive all night because I get tired but there’s no reason not to let the car drive itself overnight.

  • avatar
    Robert Schwartz

    Deal with it folks, the era of human driven cars is coming to an end. Most people I see driving are just not paying attention, they are texting, yakking on the phone, working on their make-up, reading the paper, or simply daydreaming. The autonomous car will happen, and it will happen soon. It will be a luxury at first, standard equipment later, and mandatory not long thereafter. Bet on it.

    • 0 avatar
      mcs

      it will happen soon

      No, it’s not going to happen anytime soon. The google vehicles 200k miles is misleading. I’m working on autonomous vehicle technology and already have a couple of collision avoidance systems on the market. It’s relatively easy to do what they’re doing, but the technology needed to build true autonomous vehicles is still many years away.

      Their testing methodology is designed more to produce marketing fluff and publicity rather than a true autonomous vehicle. I guess because of my collision avoidance background, I devise test scenarios from real life where I think there will be a failure, so the 200k number makes me laugh.

      The problem that still needs to be cracked with autonomous vehicles is replicating a humans knowledge of all of the objects that exist in the world and their behavior.

      For example, what about telling the difference between a cyclist, equestrian, and a moose traveling along the shoulder of a road. Passing the moose will produce an entirely different result than an equestrian or a cyclist. You have to develop technology that can determine what the hazard is and it’s behavior so the vehicle knows not to try and pass the moose, but it’s ok for the cyclist and maybe the horse.

      What about after sports events like NE Patriots games when some traffic is directed to drive in a lane left of the double yellow line on route 1? 200k miles? I’m betting their system would freak out when all of the road markings are suddenly in reverse on a North American road. What about a ball rolling into the street? Is it going to recognize the fact that if in a particular instance the ball itself isn’t a collision hazard, that an unseen child may be following and jump out between parked cars.

      It takes a lot of computing power and right now that probably means more electric power than it takes to move the vehicle down the road. You think EV range is bad now – try adding large quantities of power sucking nvidia gpus and data storage devices? We also need AI and object recognition technology that isn’t quite there yet. We have a long way to go.

      • 0 avatar
        chuckrs

        Your scenarios are situational awareness based and that is a big problem. Add that sixth sense that comes with experience that says – see that car over there, something’s wrong about how it is being handled, give it a wide berth. Happens all the time, but setting the rules that govern that sense/intuition is going to be difficult. And when do you play the odds?
        I want an autonomous flying car, please. I was promised that many decades ago….

      • 0 avatar
        wmba

        @ mcs:

        Thank goodness for a serious reply!

        If we eventually can get a car to drive itself in all conditions (my country road following the lake shore, narrow, sinuous, twisty, hilly, off camber, in blizzard conditions and no center line visible with rutted icy tracks and oncoming traffic) that team should be awarded a gold star. Nay, TWO gold stars. If the system could also dodge potholes in spring instead of ramming straight into it and ruining the wheels and suspension, well, I’d be prepared to give 3 gold stars and a Nobel nomination, because with all the variables involved, we’ll have figured out teleportation before then — it’d be easier. Four stars if it’s also an EV with 500 mile range and a great heater, that’s my limit, four stars.

        But why stop there? Hell, let’s have a household robot that does all the chores, cooks, shops, cleans, gardens, and acts as the chauffeur as well! Now that would be something if we’re pie-in-the sky dreaming – which seems to be the prevalent response here.

      • 0 avatar
        protomech

        I stopped reading at “right now that probably means more electric power than it takes to move the vehicle down the road.” .. granted, you were almost done at that point.

        At freeway speeds an electric car uses around 20 kW. If collision-avoidance tech is using 2kW+ for processing then it’s probably too expensive to sell. Yeah, the DARPA entries used a ton of processing power. Commercial-ready systems won’t.

        Also re: sucking power, a good autonomous agent can time traffic signals and use navigation / terrain maps to drive more efficiently AND safely than most hyper-milers. I would bet in light traffic that autonomous cars can use less total energy per mile than 3/4 of human-driven cars.

        The enemy of “good enough” is perfection.. can an autonomous car handle edge cases better than a human? Will the autonomous car handle stopped traffic on the interstate markedly better than a human? Will the autonomous car handle jaywalkers and children better than a human driver? If the answer is no, then the systems are not ready for use.

        The first truly autonomous agents will probably only work on the freeway. They’ll probably require a human in the driver seat ready to take over in the case of the autonomous agent alerting the driver to unusual situations (like a line of cars entering on the “wrong” side of the road). And just like with cruise control systems, the human is still responsible for operating the vehicle — unless he can show that an accident was caused by manufacturer defect.

      • 0 avatar
        Herm

        its already happening with adaptive cruise control and lane departure “warnings” that actually move your steering wheel.. its coming and its unavoidable.. thank god!

    • 0 avatar
      nonce

      Driving perfectly in all conditions is not the goal. The goal is better than humans in most conditions.

  • avatar
    forraymond

    I wish this tech was available today. I would use it most of the time, but I do still enjoy driving, just mundane drives could be better utilized working rather than avoiding bad drivers.

  • avatar
    vww12

    We own 04, 06, and 08 Euro cars we are unlikely to replace anytime soon, as these are pampered and maintained fastidiously to an as-new standard.

    However, the nanosecond self-driving cars become available, these three are getting dumped immediately. Who can stand driving 500-mile trips at 65 MPH on our cars that were made to cruise at 120+ MPH?

    With self-driving machines, at least we’ll be able to catch a movie at 65MPH.

  • avatar
    Garak

    Sorry, but completely self-driving cars will never see the light of day unless we build special, walled-off roads for them. Lawyers will take care of that. The first autonomous automobile accident – no matter how minor – will destroy the company who built the car.

    • 0 avatar
      Robert Schwartz

      Wouldn’t it be cheaper and faster to take care of the lawyers?

      Q. What do they call 7 skydiving lawyers?

      A. Skeet.

    • 0 avatar

      If American regulators will not protect manufacturers from trial lawyers they’ll simply manufacture and sell these vehicles in the world’s largest auto market instead, China. And if that is where they are being sold you may as well move the R&D there as well.

    • 0 avatar
      protomech

      If they require special walled-off roads to drive on, then they probably will never be feasible.

      Google’s autonomous agents drive on the same roads everyone else does – albeit with a human minder ready to take the wheel.

  • avatar
    wallstreet

    Firstly, a Lexus can park itself.

    Secondly, a Google Prius can navigate by itself.

    Finally, we are moving closer to the day where all auto insurers are out of business.

  • avatar
    thrashette

    Perhaps I’m just not able to wrap my itty-bitty pea brain around this technology, but I don’t understand how this could work. I don’t understand how the computer can NOT be confused by malicious individuals, or even “hacked,” so to speak. I don’t understand how people expect a COMPUTER to reliably transport them for years. Computers are not generally reliable. Computers break. Sure, if the a computer in one of OUR autonomous cars breaks, we’d be able to take over and drive our way to see a (n expensive) mechanic. But what if we’re reading or sleeping? What if it’s not manned? And what if it’s one of the people of the “future,” or a new driver who doesn’t know how to properly operate a motor vehicle? I don’t understand what’s so difficult about driving anyway. It’s fun. And when it’s not fun, like on a freeway late at night, you can hit cruise and turn the wheel a bit every now and then. Maybe I’m just easily amused. Maybe I’m just naive and stupid. But I’m not an old person who’s afraid of technology… I just would never trust a computer to navigate a 3-ton piece of steel among other 3-ton pieces of steel

    • 0 avatar
      jmo

      I don’t understand how people expect a COMPUTER to reliably transport them for years.

      The flight control computer/autopilot transports millions of people safely all over the world. The flight control computers are very reliable.

      • 0 avatar

        There aren’t nearly as many obstacles in the sky as on the ground. When my niece was 7 I took her for flying lessons. It didn’t scare me a bit. On the other hand, around the same time, I taught a woman to drive. That was quite scary.

    • 0 avatar
      protomech

      As a motorcyclist, I’d trust a good autonomous agent over the majority of cellphone-wielding human drivers on the road.

  • avatar
    MrFixit1599

    As much as I hate to admit it, I am looking forward to the day when I can set my cruise control and make my 300 mile drive through Milwaukee and Chicago to get to my customer that day. I work for a service company, and my jobs are all over the midwest. I drive about 45,000 miles a year getting to customers and back and have done so for 15 years. Never had an accident, but lord knows there have been MANY times when if it not for me paying attention, and getting out of the way of some idiot, I would have been. Again, sadly, give me cars all around me controlled by computers and let me make my own choices. In the end computers are better at making driving decisions than a good portion of the drivers on the road. All they have to do is drive, they dont put on make-up, or eat a cheeseburger, or text (not sure about that) while they are driving. The Maximum Overdrive movie does come to mind with this post, and if you haven’t seen it you are missing out if only for the soundtrack.

  • avatar
    Glenn Mercer

    Just thinking…. Would you board an airliner if there were no pilots up front? Or am I thinking like a curmudgeon or dinosaur?

    • 0 avatar
      jmo

      Would you board an airliner if there were no pilots up front?

      “Unfortunately, instead of following the established stall recovery procedure of adding full power and lowering the nose to prevent the stall, the captain only added about 75% power and continued applying nose-up inputs. As the aircraft came even closer to stalling the stick pusher activated (“The Q400 stick pusher applies an airplane-nose-down control column input to decrease the wing angle-of-attack [AOA] after an aerodynamic stall”).[21] The captain overrode the pusher and continued pulling on the control yoke resulting in the upset and subsequent loss of control.[22]”

      My money is on the computer.

      http://en.wikipedia.org/wiki/Colgan_Air_Flight_3407

  • avatar
    LeMansteve

    Clearly autonomous driving is a piece of cake. Their capabilities on today’s roads are limited partly due to some exceptions mentioned above – special traffic patterns, recognizing foreign objects on the road, collision avoidance, etc. The real challenges will be in changing the services such as police, insurance, GPS and changing infrastructures that support driving as we know it today, and changing them in a way to support autonomous cars.

  • avatar
    Herm

    human controlled cars and planes are weapons, weapons are always regulated in our society.. luckily its becoming possible now.

  • avatar
    nikita

    A modern Airbus or Boeing is fully capable of takeoff, landing and even taxi to the gate without human intervention. No one operates that way. The joke is that future airliners will have one pilot and a dog up front. The pilot’s job is to feed the dog and the dog’s job is to bite the pilot if he touches anything.

    We, as a society, still need someone to blame, and not just the engineers that programmed it, if people die. The tort side of law is simple, the owner, by its insurer, pays. The legal problem is on the criminal side if something really bad happens.

    I want an autonomous car before I reach the age where I can no longer drive.

    • 0 avatar
      DXTR

      “A modern Airbus or Boeing is fully capable of takeoff, landing and even taxi to the gate without human intervention.”

      Boeing commercial jets can not takeoff of taxi by themselves; pilots accomplish all takeoffs and taxi operations. They are capable of accomplishing an automatic landing, but not without human intervention: the aircraft must still be configured for landing (extending the flaps on schedule during deceleration, putting the landing gear down, and arming the speedbrake) and selection of the approach mode must be manually selected. If any problems are encountered during the approach and landing, the pilots must recognize them and take over manually. And problems do occur.

  • avatar
    Barry Sweezey

    Lawyers should get over themselves. “Oh, making the car autonomous is the easy part.”

  • avatar
    shwabbie

    I do not know when they will perfect the technology, laws or insurance issues… but I do know that it will be worth the wait.

    I just lost my 17 year old nephew to a car crash this weekend and after hearing that car crashes are the number one cause of death in the United States… all I know is that something needs to be done. I heard that over 90% of car accidents are ‘Operator Error’ as was the case in my nephews death… so if there is eventually a way to automate cars, I am all for it.

    btw… don’t planes have ‘Auto Pilot’???


Back to TopLeave a Reply

You must be logged in to post a comment.

Subscribe without commenting

Recent Comments

New Car Research

Get a Free Dealer Quote

Staff

  • Authors

  • Brendan McAleer, Canada
  • Marcelo De Vasconcellos, Brazil
  • Matthias Gasnier, Australia
  • W. Christian 'Mental' Ward, Abu Dhabi
  • Mark Stevenson, Canada
  • Faisal Ali Khan, India