7 ms·

So I'll spin the title a different way: "Billions are being invested in a robot that will truly liberate an aging population, the injured and the disabled, and

So I'll spin the title a different way: "Billions are being invested in a robot that will truly liberate an aging population, the injured and the disabled, and those that simply don't care to own an automobile." The poor title notwithstanding, the trust issues cited in the article are good to be aware of. Nevertheless, I'll be shocked if self-driving vehicles aren't a technology that utterly reshapes transportation in the U.S.A. in very short order.

Aelinsaar 10y ago
It would be a compelling article if humans didn't have such a dismal and fatal track record on the road.
- a3n 10y ago
  It's not entirely humans' fault, even when humans are at fault. Road implementation is a contributing factor, for example. I'm amazed that so many people in technology think that these rolling physical software boxes are going to be reliable enough to be better than us. We're pretty good at processing new, ambiguous situations. We can't even build a secure internet. Which, by the way, these cars will be part of. I think what will happen is that some accident categories will go down, others involving autonomous cars will go up, and there will be a lot of instances of people who can't get where they need to go because the car can't figure out some ambiguous situation. Or a car will try to follow a Google route that goes around a block three times before it's (hopefully) finally routed on its way. And there will be rare instances of cars trying to drive to Ireland through the Pacific Ocean.
  - dragonwriter 10y ago
    > It's not entirely humans' fault, even when humans are at fault. Yes, it is. > Road implementation is a contributing factor, for example. "Road implementation" is done by humans, so, even if it is a contributing factor, that doesn't make it any less humans' fault.
    - twinkletwinkle 10y ago
      You're intentionally mis-understanding what he wrote. He means it isn't always the human driver's fault. Driverless cars aren't going to re-implement roads (not at first, not for a while).
  - zyxley 10y ago
    > And there will be rare instances of cars trying to drive to Ireland through the Pacific Ocean. It's silly to assume that a car is going to trust a mapping service over its own sensors saying "there's no road or path here". That said, I absolutely do expect cases of "I wanted to visit a friend in the next state, fell asleep overnight, and woke up in a different town of the same name eight states away".
  - fucking_tragedy 10y ago
    The idea of software developers deciding how a 2-ton hunk of metal and plastic should whip around the road at 70mph is horrifying. Even with MISRA, we had the Toyota unintended acceleration case (not to say the developers were following it though. Grab the court transcripts if you have the time, they're brilliant). Probably dozens of other corner cases we'll never know about either.
  - Aelinsaar 10y ago
    If those accident categories are the ones that don't include fatalities or even injures (which is likely) then it's a win. Computers and computer vision systems are probably less likely to fail on a highway, than a human is. Computers can also integrate road data in ways that humans can't. To be honest, I think the argument you're offering is just the reason why we'll start seeing automation of route-following vehicles first. Trucks, buses, and taxis going to airports and hotels, don't necessarily have hard route-finding problems. You also say things like "Can't build a secure internet" as though that's ever been tried, or necessary in this case.
    - a3n 10y ago
      Never been tried? There's a green padlock next to my address field. That looks like a try. :) And OK, let's say it's never been tried. Those cars are going to be part of that internet.
      - Aelinsaar 10y ago
        I think that "looks like" is the active ingredient there.
- ars 10y ago
  Is it really so dismal? Let's run some numbers: The accident rate is around 74 per 100 million miles (and fatalities is 1.13). It's unclear exactly how to turn that into a percentage, but no matter how you do it it's quite high. Say an accident takes 5 minutes, and people drive 30 miles/hour. Then that works out to 99.999% for humans. If you use the numbers for fatalities then it's 99.99999%. So the human track record for fatalities is a failure rate of 0.00001% - driverless cars are going to have a VERY hard time beating that. Would you like to calculate what that works out to for the reliability of each individual computer, or order to have such good results in aggregate?
  - yourapostasy 10y ago
    This is an interesting analysis, but I'm not sure we have enough data on autonomous cars to know one way or the other. Perhaps failsafe protocols can help mitigate catastrophic effects from computer hardware and software failures, and that edges out humans who override equivalent "failures" like impaired driving capacity. The data on US traffic fatalities [1] is interesting, because you point out it will be hard for the autonomous vehicles to beat the average, but in some sparsely-populated states the rate is dramatically lower, which could potentially challenge autonomous vehicles' safety records even more in those states. If a lot of their safety is developed from training, and they don't train much in the sparsely-populated states because they're so low population, then maintaining the safety rate will depend a lot upon whether the training in more populous states adequately covers use cases encountered in the field in the sparsely-populated states. I suspect that at the end of the day, unless autonomous vehicles are much worse on the fatality rate, they only have to roughly match the human driver-based fatality rate, and commercial interests will push extremely hard for autonomous vehicles. The number of miles driven in the US is tracked by the Federal Reserve among others [2]. Current 12 month moving total is 3,147,579 millions of miles (3.14*10^12). Latest figure for just crashes (not only those with fatalities) is a 2011 article [3] that asserted 185 per 100 million Vehicle Miles Traveled (VMT). The Insurance Information Institute has a report [4] that cites a $277 billion in 2010 cost of crashes, not including quality of life valuations. Autonomous cars don't have to move the needle that much over the current regime before insurance will put in very strong incentives to use them; there is a ton of room to improve, and the insurance companies stand to gain a huge windfall by not lowering rates commensurately with actual cost savings. The trucking companies must be going bonkers over the possibilities to run diesel trucks 24x7. [1] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview http://www.iihs.org/iihs/topics/t/general-statistics/fatalit... [2] https://research.stlouisfed.org/fred2/series/M12MTVUSM227NFWA https://research.stlouisfed.org/fred2/series/M12MTVUSM227NFW... [3] http://www.caranddriver.com/features/safety-in-numbers-charting-traffic-safety-and-fatality-data http://www.caranddriver.com/features/safety-in-numbers-chart... [4] http://www.iii.org/issue-update/auto-crashes http://www.iii.org/issue-update/auto-crashes
atemerev 10y ago
Not much, compared to taxis and Uber already available.
- monk_e_boy 10y ago
  With both you have to pay for the humans time. Robot cars will be cheaper. Given the choice for £x a mile for a human driver or (£x * 0.2) for a robot car?
- mdonahoe 10y ago
  The dramatic changes will come when nearly cars are autonomous, and the rules of the road start changing to optimize for computers. In the meantime, I agree that you can approximate a self-driving car experience pretty closely by ordering an Uber.
  - kbenson 10y ago
    We'll probably see automated driver only roads, which allow a separate mode to be used where the AI can assume (or at least assume with much more certainty) that the other drivers are AI, and optimize accordingly. Shorter gap distances, higher speeds, etc. As long as human drivers are allowed on a particular segment of road, AI will always have to be somewhat more cautious, as the risk is still too high.
- ItsDeathball 10y ago
  The problem there is cost. Even Uber and Lyft are too pricey for most people to rely on them for a daily commute. If autonomous cars drive that cost down to something comparable to bus fare, that's going to cause a major shift in the way people get around.
  - petra 10y ago
    That doesn't count the future of UberPool , and scaling it to on-demand jitneys, etc.
  - atemerev 10y ago
the_cat_kittles 10y ago
dont forget saving an annual ~30k lives domestically (if everyone stopped driving themselves)
- TheOtherHobbes 10y ago
  As soon as one person dies in an accident that might be caused by a software error, the lawyers are going to be all over the case quicker than you can say "Astronomical compensation claim." It's bad enough now with ambulance chasers, but when the "guilty" party is a corporation worth billions, it's going to get a whole lot worse. Personally I'd love self driving cars - tomorrow, if not sooner. Most driving is boring and pointless and I'm totally happy to let a competent machine do it. But politically too many people have been conditioned to see driving as an expression of personal freedom, and the whole industry is politically vulnerable in a way that the smartphones and the Internet never were. Generally there's a whole extra layer of legal, political, social, and psychological confusion around any kind of computing that doesn't live in a box with a screen, but starts being present in and manipulating physical reality. Cars are just the start, and I think they'll turn out to be one of the more straightforward problems.
  - undersuit 10y ago
    >too many people have been conditioned to see driving as an expression of personal freedom Conditioned by some of the very same companies that are going to take away the expression, but the citizens will blame the government for this "restriction of freedom," while those same companies keep making profit regardless.
    - dragonwriter 10y ago
      > Conditioned by some of the very same companies that are going to take away the expression Those companies have been very good at various reshapings of public demand when it served their interests, I'd be very surprised if, when it suits them for automated cars to be the next thing, they aren't just as good as when they made minivans and then SUVs must-haves through marketing (when the real motivation for each was to exploit regulatory exceptions.)

Aelinsaar 10y ago

It would be a compelling article if humans didn't have such a dismal and fatal track record on the road.

a3n 10y ago
It's not entirely humans' fault, even when humans are at fault. Road implementation is a contributing factor, for example. I'm amazed that so many people in technology think that these rolling physical software boxes are going to be reliable enough to be better than us. We're pretty good at processing new, ambiguous situations. We can't even build a secure internet. Which, by the way, these cars will be part of. I think what will happen is that some accident categories will go down, others involving autonomous cars will go up, and there will be a lot of instances of people who can't get where they need to go because the car can't figure out some ambiguous situation. Or a car will try to follow a Google route that goes around a block three times before it's (hopefully) finally routed on its way. And there will be rare instances of cars trying to drive to Ireland through the Pacific Ocean.
- dragonwriter 10y ago
  > It's not entirely humans' fault, even when humans are at fault. Yes, it is. > Road implementation is a contributing factor, for example. "Road implementation" is done by humans, so, even if it is a contributing factor, that doesn't make it any less humans' fault.
  - twinkletwinkle 10y ago
    You're intentionally mis-understanding what he wrote. He means it isn't always the human driver's fault. Driverless cars aren't going to re-implement roads (not at first, not for a while).
- zyxley 10y ago
  > And there will be rare instances of cars trying to drive to Ireland through the Pacific Ocean. It's silly to assume that a car is going to trust a mapping service over its own sensors saying "there's no road or path here". That said, I absolutely do expect cases of "I wanted to visit a friend in the next state, fell asleep overnight, and woke up in a different town of the same name eight states away".
- fucking_tragedy 10y ago
  The idea of software developers deciding how a 2-ton hunk of metal and plastic should whip around the road at 70mph is horrifying. Even with MISRA, we had the Toyota unintended acceleration case (not to say the developers were following it though. Grab the court transcripts if you have the time, they're brilliant). Probably dozens of other corner cases we'll never know about either.
- Aelinsaar 10y ago
  If those accident categories are the ones that don't include fatalities or even injures (which is likely) then it's a win. Computers and computer vision systems are probably less likely to fail on a highway, than a human is. Computers can also integrate road data in ways that humans can't. To be honest, I think the argument you're offering is just the reason why we'll start seeing automation of route-following vehicles first. Trucks, buses, and taxis going to airports and hotels, don't necessarily have hard route-finding problems. You also say things like "Can't build a secure internet" as though that's ever been tried, or necessary in this case.
  - a3n 10y ago
    Never been tried? There's a green padlock next to my address field. That looks like a try. :) And OK, let's say it's never been tried. Those cars are going to be part of that internet.
    - Aelinsaar 10y ago
      I think that "looks like" is the active ingredient there.
ars 10y ago
Is it really so dismal? Let's run some numbers: The accident rate is around 74 per 100 million miles (and fatalities is 1.13). It's unclear exactly how to turn that into a percentage, but no matter how you do it it's quite high. Say an accident takes 5 minutes, and people drive 30 miles/hour. Then that works out to 99.999% for humans. If you use the numbers for fatalities then it's 99.99999%. So the human track record for fatalities is a failure rate of 0.00001% - driverless cars are going to have a VERY hard time beating that. Would you like to calculate what that works out to for the reliability of each individual computer, or order to have such good results in aggregate?
- yourapostasy 10y ago
  This is an interesting analysis, but I'm not sure we have enough data on autonomous cars to know one way or the other. Perhaps failsafe protocols can help mitigate catastrophic effects from computer hardware and software failures, and that edges out humans who override equivalent "failures" like impaired driving capacity. The data on US traffic fatalities [1] is interesting, because you point out it will be hard for the autonomous vehicles to beat the average, but in some sparsely-populated states the rate is dramatically lower, which could potentially challenge autonomous vehicles' safety records even more in those states. If a lot of their safety is developed from training, and they don't train much in the sparsely-populated states because they're so low population, then maintaining the safety rate will depend a lot upon whether the training in more populous states adequately covers use cases encountered in the field in the sparsely-populated states. I suspect that at the end of the day, unless autonomous vehicles are much worse on the fatality rate, they only have to roughly match the human driver-based fatality rate, and commercial interests will push extremely hard for autonomous vehicles. The number of miles driven in the US is tracked by the Federal Reserve among others [2]. Current 12 month moving total is 3,147,579 millions of miles (3.14*10^12). Latest figure for just crashes (not only those with fatalities) is a 2011 article [3] that asserted 185 per 100 million Vehicle Miles Traveled (VMT). The Insurance Information Institute has a report [4] that cites a $277 billion in 2010 cost of crashes, not including quality of life valuations. Autonomous cars don't have to move the needle that much over the current regime before insurance will put in very strong incentives to use them; there is a ton of room to improve, and the insurance companies stand to gain a huge windfall by not lowering rates commensurately with actual cost savings. The trucking companies must be going bonkers over the possibilities to run diesel trucks 24x7. [1] http://www.iihs.org/iihs/topics/t/general-statistics/fatalityfacts/state-by-state-overview http://www.iihs.org/iihs/topics/t/general-statistics/fatalit... [2] https://research.stlouisfed.org/fred2/series/M12MTVUSM227NFWA https://research.stlouisfed.org/fred2/series/M12MTVUSM227NFW... [3] http://www.caranddriver.com/features/safety-in-numbers-charting-traffic-safety-and-fatality-data http://www.caranddriver.com/features/safety-in-numbers-chart... [4] http://www.iii.org/issue-update/auto-crashes http://www.iii.org/issue-update/auto-crashes

atemerev 10y ago

Not much, compared to taxis and Uber already available.

monk_e_boy 10y ago
With both you have to pay for the humans time. Robot cars will be cheaper. Given the choice for £x a mile for a human driver or (£x * 0.2) for a robot car?
mdonahoe 10y ago
The dramatic changes will come when nearly cars are autonomous, and the rules of the road start changing to optimize for computers. In the meantime, I agree that you can approximate a self-driving car experience pretty closely by ordering an Uber.
- kbenson 10y ago
  We'll probably see automated driver only roads, which allow a separate mode to be used where the AI can assume (or at least assume with much more certainty) that the other drivers are AI, and optimize accordingly. Shorter gap distances, higher speeds, etc. As long as human drivers are allowed on a particular segment of road, AI will always have to be somewhat more cautious, as the risk is still too high.
ItsDeathball 10y ago
The problem there is cost. Even Uber and Lyft are too pricey for most people to rely on them for a daily commute. If autonomous cars drive that cost down to something comparable to bus fare, that's going to cause a major shift in the way people get around.
- petra 10y ago
  That doesn't count the future of UberPool , and scaling it to on-demand jitneys, etc.
- atemerev 10y ago

the_cat_kittles 10y ago

dont forget saving an annual ~30k lives domestically (if everyone stopped driving themselves)

TheOtherHobbes 10y ago
As soon as one person dies in an accident that might be caused by a software error, the lawyers are going to be all over the case quicker than you can say "Astronomical compensation claim." It's bad enough now with ambulance chasers, but when the "guilty" party is a corporation worth billions, it's going to get a whole lot worse. Personally I'd love self driving cars - tomorrow, if not sooner. Most driving is boring and pointless and I'm totally happy to let a competent machine do it. But politically too many people have been conditioned to see driving as an expression of personal freedom, and the whole industry is politically vulnerable in a way that the smartphones and the Internet never were. Generally there's a whole extra layer of legal, political, social, and psychological confusion around any kind of computing that doesn't live in a box with a screen, but starts being present in and manipulating physical reality. Cars are just the start, and I think they'll turn out to be one of the more straightforward problems.
- undersuit 10y ago
  >too many people have been conditioned to see driving as an expression of personal freedom Conditioned by some of the very same companies that are going to take away the expression, but the citizens will blame the government for this "restriction of freedom," while those same companies keep making profit regardless.
  - dragonwriter 10y ago
    > Conditioned by some of the very same companies that are going to take away the expression Those companies have been very good at various reshapings of public demand when it served their interests, I'd be very surprised if, when it suits them for automated cars to be the next thing, they aren't just as good as when they made minivans and then SUVs must-haves through marketing (when the real motivation for each was to exploit regulatory exceptions.)