back to article Rear-end crashes prompt probe into Amazon's Zoox self-driving cars

The US National Highway Traffic Safety Administration (NHTSA) is investigating two rear-end crashes involving Amazon-owned Zoox self-driving cars and motorcycles. The case was opened on Friday with an Office of Defects Investigation (ODI) preliminary evaluation that summarizes the circumstances of the two crashes, both of …

  1. captain veg Silver badge

    Self-driving

    What we actually mean is computer-driving.

    This is not necessarily a bad thing, but so far, on real roads, it clearly is.

    I propose that all the so-called self-driving vendors put their machines up against each other on a closed track. Pay unemployed actors and medical students to wander among them. Find out just how rubbish they are for avoiding wetware damage in anything other than ideal situations and ban them from any other, which includes, so far as I can tell, pretty much all the roads that ordinary people drive along.

    Should they manage this rather low hurdle there is an eminently sensible place for auto-drivers. Somewhere where all vehicles are travelling in the same direction and won't (usually) veer off unexpectedly. Motorways. As a bonus, since the vehicles follow an entirely predictable trajectory they could be recharged by, say, an overhead catenary.

    -A.

    1. AVR Bronze badge

      Re: Self-driving

      And then after leaving the arena, the self-driving cars would switch back to normal mode so they could resume collecting souls for Arioch.

    2. Throatwarbler Mangrove Silver badge
      Terminator

      Re: Self-driving

      I would very much like to see a comparison of autonomous vehicle vs. human driver accidents per kilometer driven. My guess is that the autonomous vehicles would be revealed to be orders of magnitude safer, but I'm certainly happy to be proven wrong by actual data.

      1. Dinanziame Silver badge

        Re: Self-driving

        For car deaths, we don't have enough data. Though you'd think there's a lot of car deaths, the number in the US is only 15 deaths for one billion vehicle miles. Proper self-driving cars haven't driven nearly enough to get a reliable estimate, partly because they generally don't drive on highways.

        1. John Robson Silver badge

          Re: Self-driving

          Tesla report about 3.5 billion miles... so that would be an expected death toll of 50-55.

          But deaths aren't the only measure, serious injuries are going to be more frequent, and therefore a more significant expected sample.

          Correcting for road types is difficult however.

          1. Anonymous Coward
            Anonymous Coward

            Re: Self-driving

            Tesla cars are not self driving. That's why every Tesla freeway accident is ultimately the responsibility of the driver who was supposed to be watching the road.

            1. John Robson Silver badge

              Re: Self-driving

              Correct, though they are already safer than the average human in a fairly wide range of circumstances.

              Remember the old joke about cruise control and people going to make a cup of tea in an RV...

              1. Alan Brown Silver badge

                Re: Self-driving

                This is WHY Tesla's not allowed to call it "autopilot" in Germany

                Other nations should be following that lead

                (Yes, I know an aircraft autopilot can be as dumb as "keep the wings level and fly this heading", but that's not how the general public sees them)

      2. Ian Johnston Silver badge

        Re: Self-driving

        That would certainly be interesting ... as long as the roads used were the same.

    3. tip pc Silver badge

      Re: Self-driving

      Your description sounds more like death race 2000

      https://www.youtube.com/watch?v=KB6gOFyRTN8

  2. Paul Hovnanian Silver badge

    Rear-end crashes prompt probe into motorcycles following too closely

    There. Fixed the problem for you.

    1. veti Silver badge

      Re: Rear-end crashes prompt probe into motorcycles following too closely

      Well, kinda. The bikes clearly were too close, by definition. But it's still worth investigating whether the computer driver could have handled itself better in these situations.

      1. John Robson Silver badge

        Re: Rear-end crashes prompt probe into motorcycles following too closely

        You can't not stop for a pedestrian because a motorcycle is being ridden incompetently.

        1. Paul Hovnanian Silver badge

          Re: Rear-end crashes prompt probe into motorcycles following too closely

          This.

          Problem is that many drivers (motorcycle and otherwise) try to look ahead and anticipate the general flow of traffic. Sometimes that's smart. Other times, not so much. You might be behind an automated car with reaction times far faster than a human. Or behind some moron that just dropped their hot coffee in their lap.

          You are responsible for your own stopping distance.

        2. veti Silver badge

          Re: Rear-end crashes prompt probe into motorcycles following too closely

          But were they "stopping for a pedestrian"? And if they were, could they have anticipated it earlier and made the stop more gradually?

          How will anyone ever know unless they investigate?

          If you're just going to shrug an accident off as human incompetence, then we'll never learn and never improve anything.

          1. John Robson Silver badge

            Re: Rear-end crashes prompt probe into motorcycles following too closely

            If you drive into the back of someone you're responsible for that.

            I'm sorry if my ideas relating to driving safely don't seem all that appealing.

            1. Michael Wojcik Silver badge

              Re: Rear-end crashes prompt probe into motorcycles following too closely

              I would say that there are exceptions, but they're rare. For example, if the brakes fail suddenly on your car (and the car has been maintained in a reasonable fashion), then your vehicle is to blame, but you as the driver are not. (And yes, this sort of thing does happen.)

              Another example might be an oil slick on the road, which can be difficult to see and can greatly increase stopping distance.

              I've also been in situations where it is effectively impossible to maintain a reasonable distance, because if you attempt to, other drivers will immediately pull into your lane in front of you. You can't maintain a distance of more than a car length for any significant period of time. That was common during commuting hours on the highways in the Boston area, for example (and presumably still is).

              1. John Robson Silver badge

                Re: Rear-end crashes prompt probe into motorcycles following too closely

                Vehicle failure is vanishingly rare, but yes - assuming a reasonable maintenance log then that would be not a personal fault. I would suggest that it doesn't happen often in most cases - runaway trucks being the exception...

                An oil slick on the road will, in theory, prevent the vehicle in front from stopping as well - assuming they aren't putting the oil down of course.

                People cutting you up isn't a reason to close that distance, it's frustrating, but you can always maintain a safe braking distance - though that does come to the true exception... When someone pulls in front of you and slams on their brakes, and the similar - when someone has no functioning brake lights (which is why indicators are different lights USians) and does the same. Both of those are more common deliberately than they are "incidentally".

                1. Alan Brown Silver badge

                  Re: Rear-end crashes prompt probe into motorcycles following too closely

                  "runaway trucks being the exception..."

                  Runaway trucks are almost always a result of poor driving technique resulting in brake overheating when the driver should have been using the engine for braking. The remaining few instances are poor maintenance

                  If these happen the authorities are usually all over the owners and driver like a badly fitting shirt to find the cause before someone gets killed

              2. Alan Brown Silver badge

                Re: Rear-end crashes prompt probe into motorcycles following too closely

                "effectively impossible to maintain a reasonable distance, because if you attempt to, other drivers will immediately pull into your lane in front of you"

                Common throughout the world and a good time to hover your left foot over the brake even if you do have automated braking

                The old joke is that most cars know the laws of Newtonian physics better than most drivers and will behave accordingly if asked to break them

                The only real basis the NHTSA has for investigation is "sudden stop - for no reason". Everything else is on the riders and I'd like to see the camera footage from the incidents

      2. Alan Brown Silver badge

        Re: Rear-end crashes prompt probe into motorcycles following too closely

        Some of the videos of US riders I've seen recently are flat out scary

        Lane filtering at a differential over 50mph is flat out dangerous and it's clear many US riders got their licenses from the same wheaties packets as US cage drivers

    2. Anonymous Coward
      Anonymous Coward

      Re: Rear-end crashes prompt probe into motorcycles following too closely

      I've seen this, long before self-driving cars. A motorcyclist was too close behind a car, the car hit the brakes (stopping at a light but a little suddenly), the motorcyclist couldn't stop in time and rear-ended the car. Thankfully, he was moving pretty slow on impact, so the front tire hit and the back tire jumped slightly, and that was it - no damage or injuries.

      There's a REASON we're taught to leave a certain distance between us and the vehicle in front of us!

  3. martinusher Silver badge

    You don't need self-driving to get into trouble

    As any motorcyclist will tell you the problem is really anti-lock brakes on modern cars. The driver doesn't need to think about skidding, you just pump the pedal and the car stops. More often than not it stops far quicker than you can so factoring in reaction time and you're at worst going to rear end the car in front or at best get a nasty scare. It doesn't help that most states don't allow lane splitting, they expect bikes to travel on the same path as cars, so you can't easily keep to one side or another.

    The only way to survive on a bike (apart from keeping away from crowded traffic) is to not look at the car in front but look beyond it and hopefully anticipate what its going to do. (This assumes that you are following a car and not one of those popular urban assault vehicles.) You also need to position yourself so should the worst happen you're not the meat in a tin sandwich.

    1. very angry man

      Re: You don't need self-driving to get into trouble

      Give us back our guns and it will be sorted really quickly

    2. biddibiddibiddibiddi Bronze badge

      Re: You don't need self-driving to get into trouble

      Safety by not staring solely at the one in front of you applies to every driver, and bicyclist, and pedestrian. Watch multiple cars ahead of you for signs that something is shifting and you'll have a better chance of knowing that the one in front is about to change in some way, even if you don't see any brake lights. By law, the vehicle in the rear is responsible for being able to stop safely even if the one in front comes to a sudden dead stop with no warning. The majority of us don't follow that rule because we're in a rush, have a natural tendency to think getting closer is getting there faster and not getting "left behind", and unfortunately because other people think an open space that is long enough for safety is a big welcome sign to change lanes into that space and then act stupid themselves.

      Lane splitting is just asking to die as far as I'm concerned. Jamming your motorcycle into a space where nobody in the surrounding 2 ton death machines expects an unprotected human body to be. Just because you don't want to wait like the rest of us.

      1. jmch Silver badge
        Mushroom

        Re: You don't need self-driving to get into trouble

        "...other people think an open space that is long enough for safety is a big welcome sign to change lanes into that space..."

        Aaaaarrghhhh!!!! A thousand times this!!!!

      2. DancesWithPoultry
        IT Angle

        Re: You don't need self-driving to get into trouble

        > Lane splitting is just asking to die as far as I'm concerned

        'Filtering' old boy.

        Both perfectly legal in Blighty, quite commonplace and quite safe when done properly.

        Granted, driver competence and road safety is much better on this side of the pond, however you choose to measure it: https://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

    3. Anonymous Coward
      Anonymous Coward

      Re: You don't need self-driving to get into trouble

      "As any motorcyclist will tell you the problem is really anti-lock brakes on modern cars. " Being a car driver observing how motorcyclists ride, I can tell you that the problem is not technical.

    4. Anonymous Coward
      Anonymous Coward

      Re: You don't need self-driving to get into trouble

      "As any motorcyclist will tell you the problem is really anti-lock brakes on modern cars"

      If the vehicle in front breaks for a legitimate reason and you crash into it, it's your fault 100% (at least in the UK). Blaming the other vehicle for "stopping to fast" is ridiculous. You should have left more space between you and them.

      1. ICL1900-G3 Silver badge
        Headmaster

        Re: You don't need self-driving to get into trouble

        Breaks what? Or did you mean 'brakes'?

      2. tiggity Silver badge

        Re: You don't need self-driving to get into trouble

        @AC

        Indeed, in the UK, various insurance scammers have managed to make good money from doing this, slam on brakes, get rear ended then lots of big financial claims for whiplash etc.*

        * The scammers are not stupid, they do this when car behind travelling relatively slowly, so its a relatively low kinetic energy impact, as they do not want any genuine injuries from the collision!

        1. Alan Brown Silver badge

          Re: You don't need self-driving to get into trouble

          Less so now that dashcams are commonplace...

    5. UnknownUnknown

      Re: You don't need self-driving to get into trouble

      Not weaving in and out of traffic like a maniac is also suggested.

    6. jmch Silver badge

      Re: You don't need self-driving to get into trouble

      "As any motorcyclist will tell you the problem is really anti-lock brakes on modern cars...."

      This motorcyclist is telling you it's not just ABS, it's all brakes. For every single car, even with the shittiest brakes, if you stomp on them the brakes lock and the car skids. What this tells you is that the limiting condition for optimal braking isn't the power of the brakes, it's the friction between tyre and road, which is proportional to the tyre surface area. For cars, you don't only have 2 more tyres, you have each tyre, having at least 5X the contact area as a bike tyre (since bike tyres are rounded, even quite fat bike tyres don't have that large of a surface area). Most of a bikes' stopping power is anyway on the front wheel, which is usually skinnier. So a car that weighs around 6 times a bike+rider has at least 10X stopping power. Pretty much any car can stop faster than pretty much any bike. This information is critical for bikers to realise that they need to leave a bigger gap to the car in front than if they were driving a car. Unfortunately quite a few bikers don't know and/or respect that.

      On acceleration it's different, it's only quite powerful cars / bikes that have so much power that it's more than the tyre traction can handle, resulting in wheelspin (or traction control kicking in). The limit on acceleration is usually engine power vs mass, not traction available. Even small-ish bikes have engine powers of at least half that of a quite normal car, but with around 6X less mass

      Re lane splitting, even where it's allowed, I only do it when traffic is very slow or stopped, and even then keep a safe speed differential and have to always be very aware of cars trying to change lanes without warning. In normal driving, drive to one side of the preceding car, so your tyres are running behind theirs. This allows the option of steering past if you can't brake on time. Plus, many roads are built so that if there are any manholes or other irregularities they are in the middle of the lane (where cars would go over them without touching), so driving a bit to one side will also avoid these extra bumps.

      1. Ian Johnston Silver badge

        Re: You don't need self-driving to get into trouble

        it's the friction between tyre and road, which is proportional to the tyre surface area

        Where does "surface area" appear in "F = mu W"?

        1. jmch Silver badge

          Re: You don't need self-driving to get into trouble

          "Where does "surface area" appear in "F = mu W"?"

          Very good point, and indeed in an idealised "spherical cow" world, friction (and therefore traction) depends on weight and is independent of surface area. As I understand it, the difference comes from the 'mu' part (coefficient of friction) not really being a constant over the whole contact area, or even over time, even though it is treated as such - tiny differences in road and tyre surface, differences in tyre temperature etc can change the friction coefficient.

          Even so, on further thought, maybe these differences are too small to have a material impact on car vs motorcycle stopping distances. Thanks for the correction!!

          1. Anonymous Coward
            Anonymous Coward

            Re: You don't need self-driving to get into trouble

            I'd expect the stopping distance for a motorbike to be limited by either the same factors as a car (road and tyre surface conditions), but with the added problem that in good dry braking conditions, with good tyres, you can flip yourself over the handlebars- if you can brake without going over the handlebars, the physics of braking is pretty similar for both a car and a motorbike. (i.e. leave a two second gap between yourself and the vehicle in front, and you should be OK so long as you are attentive. Admittedly, drivers / riders who obey that rule are pretty rare.... but it doesn't stop it being true.)

          2. Ian Johnston Silver badge

            Re: You don't need self-driving to get into trouble

            You're welcome! I presume there will be some non-linearity, but it has long struck m that motorcycles have remarkably small contact areas, with their semi-circular cross section tyres, compared to cars.

            1. Anonymous Coward
              Anonymous Coward

              Re: You don't need self-driving to get into trouble

              > You're welcome! I presume there will be some non-linearity, but it has long struck m that motorcycles have remarkably small contact areas, with their semi-circular cross section tyres, compared to cars.

              In theory, bikes have half the number of tyres as cars but less than half the weight (momentum) so should be able to stop better.

              In practice, under heavy braking, the front dives and the back lifts so it is the front tyre that is providing most of the stopping power.

              1. John Brown (no body) Silver badge
                Thumb Up

                Re: You don't need self-driving to get into trouble

                Yes, on a similar note, an HGV has many more, and wider tyres than a car so using just that means an HGV should stop more easily than a car. But mass just called and said "hold my beer" :-)

                (And jack-knifing with HGVs too. Every vehicle type has it's peculiarities)

        2. PRR Silver badge

          Re: You don't need self-driving to get into trouble

          > Where does "surface area" appear in "F = mu W"?

          +1

          But TRACTION (rubber on road) is not pure friction. Neither is friction. Traction is like gearing with weak teeth. Contact area has only slight effect on max force. Enough that fat tires help. But double the rubber is NO where near double the braking force. Less than square-root of patch area, maybe much-less.

          ABS: When I got my motocycle endorsement, 47+ years ago, we were told that a good motorcycle rider could always stop faster than most cars. They didn't say it this way, but motorcycles could have "wetware ABS": the rider could modulate lever and pedal just to the edge of a skid AND NO FURTHER because dropping the bike hurts bad; OTOH in 1976 cards had hard tread for good life and drum brakes which were prone to lockup and F/R unbalance; also most car drivers never practice hard braking and never learn when forced.

          You are right. My 2002 Accord Sedan did stop less-fuss and mildly shorter than my 1967 BSA 250cc.

          1. jmch Silver badge

            Re: You don't need self-driving to get into trouble

            "motorcycles could have "wetware ABS": the rider could modulate lever and pedal just to the edge of a skid AND NO FURTHER because dropping the bike hurts bad"

            I would add.... most of the bike's stopping power is on the front brake, which is hand-activated, and humans have more sensitivity in their fingers/hands than their feet (especially feet typically enclosed by shoes!). I'm sure that would help from the "wetware ABS" point of view.

            1. Alan Brown Silver badge

              Re: You don't need self-driving to get into trouble

              Most motorcycles sold this century have ABS as standard and it's been mandatory in many countries since about 2010

              I can tell you from (painful) experience that a locked up front wheel is nearly impossible to recover from before the bike disappears out from underneath you

      2. ICL1900-G3 Silver badge

        Re: You don't need self-driving to get into trouble

        Excellent post, thank you.

    7. Ian Johnston Silver badge

      Re: You don't need self-driving to get into trouble

      The only way to survive on a bike (apart from keeping away from crowded traffic) is to not look at the car in front but look beyond it and hopefully anticipate what its going to do.

      How about not riding so close to the car in front tat you couldn't stop if it braked as hard as you know it can? Is that ruled out for some reason?

      I live next to the A75 which is plagued by (a) accidents involving HGVs and (b) HGVs barrelling along on the limiter (and therefore 40% over the speed limit) a tractor unit's length from the one in front. It appears that neither they nor motorbikes are exempt from the laws of physics.

    8. Alan Brown Silver badge

      Re: You don't need self-driving to get into trouble

      "not look at the car in front but look beyond it and hopefully anticipate what its going to do"

      If you're fixating on the car in front, you're a fatal crash waiting to happen and you wouldn't get a license in most countries

  4. Anonymous Coward
    Anonymous Coward

    Easily solved

    Get rid of the two wheeled menaces.

    Or just insurance tax the whazoo out of them.

    1. Ian Johnston Silver badge

      Re: Easily solved

      I think they are dying out anyway. When I was a student in the 80s lots of my contemporaries had motorbikes because they were too skint for cars, but when you see bikers with their helmets off now they are almost all elderly people. Nobody under about 50 rides a motorbike any more, though twenty-somethings delivering food on illegal electric bikes seem to be keeping up the death rates at that end of the age range.

      1. Korev Silver badge
        Joke

        Re: Easily solved

        > Nobody under about 50 rides a motorbike any more

        A shame, we should encourage more young motorcyclists as their organs are likely to be better quality for transplanting...

        1. IHateWearingATie

          Re: Easily solved

          You're not an economist are you? :)

        2. DancesWithPoultry
          Go

          Re: Easily solved

          Wait!

          I get to have fun, look cool and be an organ donor?

          Sounds like a win-win to me!

      2. James Wilson

        Re: Easily solved

        I'll have you know I'm a motorcyclist and I'm not "about 50". I'm late 40s, dammit.

    2. Alan Brown Silver badge

      Re: Easily solved

      "Or just insurance tax the whazoo out of them."

      New Zealand operates a state-administered 3rd-party injury insurance system (ACC) which sets levies based on the costs per sector

      This levy is added to annual registration fees for vehicles (the actual registration charge is the same for all vehicles)

      It costs FIVE TIMES as much to register a motorcycle as a car there - and motorcycle levies are still substantially lower than they should be

      All things being equal, if Motorcycle ACC levies weren't being subsidised, annual registration would cost at least 20 times more than cars

      (Motorcyclists are 21 times more likely to be killed or injured on the road than cage drivers and those injuries tend to be more expensive to treat, this is despite mandatory helmet laws and widespread use of safety gear by riders)

  5. tony72

    "[...] involving Amazon-owned Zoox self-driving cars and motorcycles."

    Okay, I initially read that to mean "self-driving cars and self-driving motorcycles", I'm glad the comments set me straight, self-driving motorcycles are a bit of a scary concept.

    1. GDM
      Alert

      Do a YouTube search for self-balancing motorcycle and you'll find there's a lot of progress on this already.

  6. xyz Silver badge

    Meh...

    The push me pull you thing probably doesn't know its arse from its elbow or...

    Maybe these are cash for crash jobs. They still happen.

  7. I ain't Spartacus Gold badge
    Boffin

    I'm surprised nobody has said this before, it's so obvious

    It's the Gathering!

    There can be only one!

    It's a Toyota Highlander, after all. What else could it mean? It'll fight all the other self-driving cars until there's a single winner.

    1. Michael Wojcik Silver badge

      Re: I'm surprised nobody has said this before, it's so obvious

      Hmm. That would explain why my self-driving Hispano-Suiza has a Scottish accent.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like