Slashdot Mirror

← Back to Stories (view on slashdot.org)

Waymo Starts To Eclipse Uber in Race To Self-Driving Taxis (sfchronicle.com)

Posted by msmash on from the how-about-that dept.

Uber barreled into autonomous driving out of fear that it could end up as the MySpace or Yahoo of ride-hailing, a company with early gargantuan success that stumbled as times changed. Waymo, the self-driving offshoot of Google parent Alphabet, has pursued its ambitions more cautiously, accumulating long years of research and testing before pursuing a plan to bring its technology to the public. From a report: Now, as Waymo scales up its self-driving taxi service, Uber's fear could be coming to pass. This week, as Uber continued to reel from a fatal self-driving accident in Arizona, Waymo confidently pushed forward -- landing a deal to build 20,000 self-driving luxury SUVs with Jaguar Land Rover on top of its plan for thousands of Chysler hybrid minivans. Within two years, it aims to have thousands of fully autonomous taxis -- with no backup drivers behind the wheel -- on the roads, starting in Phoenix where it is already giving test rides.

The company predicts it will give 1 million robot-taxi rides a day by 2020. Waymo, the industry pioneer, logged millions of autonomous miles as it perfected self-driving technology. But over the years, engineers defected out of frustration that it was not commercializing the technology. Now with former auto executive John Krafcik at the helm, Waymo appears poised to launch a self-driving taxi service that could conceivably dominate that field, at least early on, the way Uber does now with human-driven cars.

43 of 67 comments (clear)

  1. Emergency STOP button by b0s0z0ku · · Score: 1

    As with any other form of dangerous machinery, I hope it at least comes with one of these...

    https://i.stack.imgur.com/jrVP...
    https://incompliancemag.com/wp...

    1. Re:Emergency STOP button by b0s0z0ku · · Score: 1

      Beat up the car for not moving, just like people used to beat dead horses?

  2. No one is close by fluffernutter · · Score: 3, Insightful

    Waymo probably has their LIDAR working, but I don't imagine that gets them very far ahead of Uber at this point. Humans drive 551,370 miles in all situations and all kinds of weather and road services without even a fender-bender. Even Waymo is doing a fraction of that per 'safety driver' intervention. Also it is a problem set that gets exponentially harder as you get closer to the goal of being significantly safer than humans.

    --
    Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    1. Re:No one is close by fluffernutter · · Score: 1

      road surfaces*

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    2. Re:No one is close by ShanghaiBill · · Score: 1

      I think this is why they are starting in Phoenix. The roads are in good condition (no freeze-thaw cycle) and the weather is almost never an issue. They already have all the roads mapped out. If your origin or destination is off the pre-mapped roads, they can dispatch a human driven car instead.

    3. Re: No one is close by fluffernutter · · Score: 4, Informative

      That *is* actual statistics. Humans drive 3.22 trillion miles in the US per year, and get into 16,000 accidents per month. It comes to 551,370 miles without an accident. We should assume each 'intervention' by a safety driver would have been an accident (the industry picked this method of measurement, not me), otherwise there wouldn't have been an intervention. Waymo is at around 4,900 miles per intervention which is 0.89% as safe as a human mile for mile; and consider that Waymo is picking *exactly* when they drive. And someone has already died. Self driving is a long way off.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    4. Re: No one is close by mileshigh · · Score: 1

      The human accident rate is actually around one per 200K miles. See my other post.

      BUT that human rate is for _actual_ accidents -- not near misses. I'd guesstimate that near misses vs. accident is better than 10:1. Let's call it 1 near miss per 20K miles. Meaning a "phew!" moment every year or two for the average driver, which feels about right.

      As far as real-world consequences, it'd be much more realistic to chalk up each intervention as a near miss. By that count, 5K miles/intervention is only 4x worse than humans.

      Room/need for improvement? Yes!!! But not decades of development.

    5. Re: No one is close by iggymanz · · Score: 1

      sorry, throw out the truck drivers and redo it.

    6. Re: No one is close by RhettLivingston · · Score: 2

      Virtually no cars go over 1/2 million miles. I just examined the parking lot outside and over half had signs of an accident of some sort. Several had obviously been in a few. The stats are way off.

      Most accidents by human drivers are unreported. As long as the damage is minor, there is no reason to raise insurance rates. EVERY accident whether or not there is any physical damage is reported for these vehicles.

    7. Re: No one is close by mentil · · Score: 1

      If the system disengages, requiring an intervention, due to a bag blowing in the road, we shouldn't assume a serious collision would otherwise occur (from the point of view of the car; the bag might disagree.) If a disengagement is accompanied by automatic braking, and/or pulling over, then there may be little risk in some situations.

      --
      Corruption is convincing someone that the selfless ideal is the same as their selfish ideal.
    8. Re: No one is close by Junta · · Score: 1

      Would be interesting the percentage of urgent interventions versus non urgent.

      I imagine an intervention can be scenarios like car not slowing down despite a pedestrian clearly about to get into the roadway, which would trigger something rather urgent and would indicate at least an increased likelihood of an accident.

      I am also imagining scenarios like the coast is clear, but the car refuses to execute some maneuver. Then the human driver intervenes and does that left turn or whatever.

      Either way, it speaks to not being able to be a taxi, at least not for people who can't drive. While they may be confident based on some histrocial imporvement in intervention rates, this is the sort of statistic that tends to come down quickly at first, and then plateauing short of your goal and eternally stuck at some unacceptable number.

      I personally am worried about the "never allow your customers to own anything ever" philosophy of the tech industry as of late, so "autonomous driving as only for taxi" greatly bothers me anyway.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    9. Re: No one is close by Junta · · Score: 1

      Conversely, throw out older cars. Of the accidents I've seen actually occur, one of them was because an axle snapped and the car completely lost it. If measuring by serious injury, only twice in my experience was an accident so severe it warranted an ambulance for anyone, and that's because that car was older than airbags (the other was not wearing a seatbelt). Also throw out bad weather conditions,country and mountain roads, etc.

      Modern cars are increasingly having things like collision alert and lane departure alerts. It is possible that a system that requires the driver to be fully engaged to operate, but also always double checking the human driver is the safest of all the options. It sucks because I'd love to be carted around, but it is an eventuality I must be prepared for.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    10. Re: No one is close by Junta · · Score: 1

      Of course, this could make for an unacceptable taxi experience. Your taxi gives up a noticable period of time isn't going to be a particularly happy experience. If it otherwise works as advertised, it might make for a solid mode for your personal car or a rental, but not really 'taxi'.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    11. Re: No one is close by fluffernutter · · Score: 1

      IF the car doesn't know what is going on, would that not cause an accident?

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    12. Re: No one is close by fluffernutter · · Score: 1

      If the car is that confused, it 'could' get in an accident simply by a bag blowing in the road. That's that point. If it understands the bag, then it is safe.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    13. Re:No one is close by fluffernutter · · Score: 1

      Ok well not one update has been much better to date, because they aren't much good at all.. when are those coming then?

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    14. Re:No one is close by MrL0G1C · · Score: 1

      "Waymo Starts To Eclipse Uber"

      That's a joke, Uber is years behind Waymo, Waymo cars get on average 5600 miles before the human driver has to intervene, Uber cars? 13 miles!!! That's a roughly 400fold difference.

      --
      Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
    15. Re: No one is close by iggymanz · · Score: 1

      plenty of weather caused accidents really are the driver going "too fast for conditions."

      My family and their acquaintances have people that love old cars and drive them, 30-60 years old vehicles, and never heard of axle breaking. I've seen that on newer cost reduced crap cars though....

  3. They've always been ahead. by Prien715 · · Score: 4, Insightful

    To get a sense of how far ahead Waymo is, take a look at the disengagements report Waymo published last year and then look at everyone else's. At least in California, Waymo has more miles driven than everyone else combined and their disengagement rate is much lower

    --
    -- Political fascism requires a Fuhrer.
    1. Re:They've always been ahead. by AvitarX · · Score: 1

      Wow, only Chevy looks close, and they're about a year behind.

      --
      Wow, sent an e-mail as suggested when clicking on "use classic" banner, and got a fast response that addressed my msg
    2. Re:They've always been ahead. by fluffernutter · · Score: 1

      What is a 'Disengagement'? That can't be the same as an 'intervention'. All accounts have Waymo at around 4900 miles per required interaction. Yet in this report there was one 'disengagement' for the entire month of Nov 2017 after driving 30,000 miles? Something doesn't add up. Also, miles driven means nothing if nothing unexpected is happening during most of those miles.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    3. Re:They've always been ahead. by swillden · · Score: 1

      Also, miles driven means nothing if nothing unexpected is happening during most of those miles.

      This depends both on what happens and on what range of events are expected/handled by the system. Most of Waymo's driving in California is on busy surface streets where lots of random things happen.

      --
      Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
    4. Re:They've always been ahead. by larryjoe · · Score: 1

      To get a sense of how far ahead Waymo is, take a look at the disengagements report Waymo published last year and then look at everyone else's. At least in California, Waymo has more miles driven than everyone else combined and their disengagement rate is much lower

      It's hard to derive any useful conclusion solely from these numbers. It's not clear what the apples-to-apples bases of comparison are. Disengagements are self-reported and perhaps interpreted differently each company. Waymo says, "The vast majority of disengagements are not related to safety." Nvidia says, "Disengagements were when a driver completed a test, or assumed manual control of the vehicle due to discomfort." So, it is not clear whether the metric is uniform or even useful for evaluation of safety.

      Second, as anyone who has driven behind a Google/Waymo car in Mountain View knows, those cars drive very conservatively, even to the point of annoying surrounding human drivers. It's not clear that the miles driven by the set of companies is comparable. A more aggressive set of driving scenarios and conditions would be more helpful for engineering feedback, but a less aggressive set of workloads would be better for marketing.

    5. Re:They've always been ahead. by R3d+M3rcury · · Score: 1

      It looks like it is. According to the CA DMV, disengagement is “a deactivation of the autonomous mode when a failure of the autonomous technology is detected or when the safe operation of the vehicle requires that the autonomous vehicle test driver disengage the autonomous mode and take immediate manual control of the vehicle.”

      It may be a terminology issue. If you're looking at Arizona, they may have different terms than California does. And whose accounts have Waymo at about 4900 miles per required interaction? Can you give us a reference?

    6. Re:They've always been ahead. by R3d+M3rcury · · Score: 1

      Whoops! Okay, I see the discrepancy. You have to figure that your statistic of 4900 miles is an average--how many miles driven divided by the number of times a person had to step in and take control of the car.

      So, yes, November 2017 had one case in 30,516.7 miles, which is pretty darned impressive. Conversely, April 2017 had 10 such disengagements in only 27,238.7 miles. Some simple math tells us that's 1 disengagement every 2,723.87 miles.

      If you take all the disengagements and divide by the total miles, you come up with about 5,596 miles per disengagement, which is pretty close to your number.

  4. Turn the damn Lidar ON next time. by truckaxle · · Score: 2

    Waymo jumps ahead... that is what happens when your competition runs someone over. Now UBER stands for UBER = >U BEtterR get out of the way!

  5. I Wouldn't Get In A Self-Driving Uber After They by dryriver · · Score: 1

    A) Killled a poor woman just because they were eager to put not-read-for-action-at-all vehicles on public roads B) Paid/settled the family of said woman into silence C) Ruined make-a-living-through-cab-driving for hundreds of thousands of low-income cabbies in dozens of developing countries with a fucking smartphone app.

    --
    Why did the chicken cross the road? Because Elon Musk put an AI chip in its head.
  6. Re:I Wouldn't Get In A Self-Driving Uber After The by 110010001000 · · Score: 1

    But I would gladly get into a Google car because Google is so benevolent.

  7. Re:I Wouldn't Get In A Self-Driving Uber After The by dryriver · · Score: 1

    Google may not be the Mother Teresa of tech objectively, but compared to Uber, they almost are Mother Teresa. And at least Google is capable of developing serious tech and has a track record of doing so. Having watched the video, I struggle to understand how Uber put a car this incapable of avoiding pedestrians on a public road at all. If you or I did that, we'd likely spend the rest of our lives in jail.

    --
    Why did the chicken cross the road? Because Elon Musk put an AI chip in its head.
  8. Re:I Wouldn't Get In A Self-Driving Uber After The by vtcodger · · Score: 1

    The issue presumably is competence, not character.

    --
    You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey
  9. Place Your Bets! by kackle · · Score: 1

    I expect these vehicles to slow down traffic all over the place as they get befuddled by road obstacles that any child could understand. But now that there's (sadly) been a death, when will Waymo cause its first?

    1. Re:Place Your Bets! by mentil · · Score: 1

      When I was in a car as a child, I'd play "stick my hand out the window and use it as a wing", "jump Mario over street cracks" and "hop Mario over the wiper blades". Not sure you want to use children as your standard for 'understanding obstacles'.

      --
      Corruption is convincing someone that the selfless ideal is the same as their selfish ideal.
  10. Numbers don't pass the smell test by mileshigh · · Score: 1

    550K miles doesn't even begin to cover the fender-bender rate. You're quoting a number similar to *police-reported* accidents. According to the DOT "In 2016, there were an estimated 7,277,000 police-reported motor vehicle crashes in the United States, resulting in 37,461 fatalities and 3,144,000 people injured." That's roughly 440K miles per _police-reported_ crash, not 550K.

    Problem is most accidents aren't police reported, and certainly not fender benders. Apparently, the average driver has an accident every 18 years. 18 x 12K miles is 200K miles.

    Think about this: 550K miles is roughly a lifetime of driving for most people, yet I don't know anyone who's never had an accident of any sort. With an average accident rate of 550K miles as you state, accident-free driving would be routine. Yet, commercial drivers get safety awards and bonuses for 500K accident-free miles. Let's just say that many pro drivers never see that bonus money.

  11. Re:I Wouldn't Get In A Self-Driving Uber After The by 110010001000 · · Score: 1

    Definitely. They aren't sucking all of your personal information and selling it to others while shoving ads in your face. They are serious tech.

  12. Simple, or stupid by SuperKendall · · Score: 1

    So, yes, November 2017 had one case in 30,516.7 miles, which is pretty darned impressive. Conversely, April 2017 had 10 such disengagements

    Well then it seems like the software is obviously getting better as we know they would have deployed a lot of updates to the system in that time, so surely the stupidest possible thing you could do is average these two points into an absolute number because it's obviously increasing in quality continuously...

    Some simple math...If you take all the disengagements and divide by the total miles,

    EPIC FACE PALM!!!!!!!!!

    I continue to be amused that Slashdot, supposedly a realm of technically astute people, cannot and will not understand how fast true self driving cars are coming. Only the most basic technical understanding of the technologies involved is required to understand this is and will happen. But here on Slashdot, it continues to be denied vehemently.

    --
    "There is more worth loving than we have strength to love." - Brian Jay Stanley
    1. Re:Simple, or stupid by Junta · · Score: 1

      I would say the data is too noisy to be that confident the trend is precisely as described. For example August looked pretty good, then september was suddenly way worse. February looked as good as October, but April was horrific.

      It is nice to review the data in more detail. It affirms that it's pretty solid on highways, less so on streets. It also shows that the most prominent intervention is user taking over due to an unwanted maneuver by the car. It shows that early on, almost all the events were human taking over without the system, and August/September had an interesting run of the software disengaging itself for some reason.

      --
      XML is like violence. If it doesn't solve the problem, use more.
  13. Re:I Wouldn't Get In A Self-Driving Uber After The by angel'o'sphere · · Score: 1

    I wonder more why they put a "driver" behind the wheel who obviously was neither qualified nor paying attention.

    --
    Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
  14. Waymo is not a Taxi service by neoRUR · · Score: 1

    So what are their plans, since they are not a taxi service and don't have the stuff Uber has set up around the world?
    So they are going to licence the cars to Uber? Sell the Tech to the highest bidder or Lyft or buyout Uber or Lyft?

    Even if they are ahead, they have no customer.

  15. Re:I Wouldn't Get In A Self-Driving Uber After The by mentil · · Score: 1

    They can just rebrand as 'Death Race 2000: The Ride' and it'll be a smash success. Throw in an AR HUD, microtransactions, and it'll be a real hit.
    Actually, Carmageddon might be a better license; kill off competing Uber cabbies to increase your surge pricing profits.

    --
    Corruption is convincing someone that the selfless ideal is the same as their selfish ideal.
  16. What might be the relative safety of such systems? by az-saguaro · · Score: 1

    This subject brings to mind a question. My post is not about the pros or cons of self driving vehicles or taxi services, not about Waymo versus Uber versus any other. It is an abstract speculation.

    Will self driving cars have different types of accidents or injury rates than human driven vehicles? For instance, if there are crashes with injuries, will the ratio of pedestrian versus occupant injuries or fatalities differ between self-driven versus human driven cars, or between different self driving companies or technologies? Will the occupants be safer than ordinary cars, and pedestrians more vulnerable, or vice versa, or no difference? Will property damages likewise be different?

    It would take an unfortunate many accidents to get such statistics, but such numbers might ultimately reveal important differences in one proprietary platform versus another. They could show which platform or technology would require better oversight or "retooling".

    This question is purely speculative for now. My question is, for those of you that are knowledgeable about the technologies, are there any insights about how car crashes or injuries might differ between human versus self driven vehicles?

  17. rickrolled by Reverend+Green · · Score: 1

    hack..
    hack..
    hack...

    waymo.all().entertainment.search('Never gonna give you up').play();
    waymo.all().navigation.left(100);
    waymo.all().navigation.accelerate(100);

    And hilarity ensues.

  18. It will be just another google 'app' on your phone by lamer01 · · Score: 1

    I mean, since they don't need drivers, they only need the cars and a place to park them during off peak hours. The whole business becomes a lot simpler if you do not need to hire drivers.

  19. Intervention != accident by gentryx · · Score: 1

    AFAIK an intervention is not an event that would have been an accident, but rather a situation in which the control software of the vehicle decides that it cannot solve the current driving situation. Without human intervention such a car is expected to pull over in a safe manner. Also, self-driving car companies are offering the combination of software + human operators for these interventions. Hence the measure should be the same as for human drivers: miles driven per accident caused.

    --
    Computer simulation made easy -- LibGeoDecomp