Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
Transportation

Self-Driving Uber Car Kills Arizona Woman in First Fatal Crash Involving Pedestrian (gizmodo.com) 953

Joe_Dragon writes: Last night a woman was struck by an autonomous Uber vehicle in Tempe, Arizona. She later died of her injuries in the hospital. The deadly collision -- reported by ABC15 and later confirmed to Gizmodo by Uber and Tempe police -- took place around 10PM at the intersection Mill Avenue and Curry Road. Autonomous vehicle developers often test drive at night, during storms, and other challenging conditions to help their vehicles learn to navigate in a variety of environments.

According to Tempe PD, the car was in autonomous mode at the time of the incident, with a vehicle operator sitting behind the wheel. A police spokesperson added in a statement that the woman's 'next of kin has not been notified yet so her name is not being released at this time. Uber is assisting and this is still an active investigation.' The woman was crossing the street outside a crosswalk when she was hit, the spokesperson said.
Update: Uber says it is suspending self-driving car tests in all North American cities after a fatal accident.
This discussion has been archived. No new comments can be posted.

Self-Driving Uber Car Kills Arizona Woman in First Fatal Crash Involving Pedestrian

Comments Filter:
  • More to come (Score:2, Informative)

    I'm very familiar with the types of sensors used in these cars. They have trouble with rain, snow, sunlight, scratches, dirt, basically anything. Try sticking your head out the window while driving and pretend your eyeballs are the sensors. It's a pretty comparable comparison. We're 50 years out from a working self-driving car. The AI isn't there, the sensors aren't there, and every amaaaaazing show-off event being held is on a perfect road with perfect weather or some conditions they know the car can hand
    • Re:More to come (Score:5, Insightful)

      by PeeAitchPee ( 712652 ) on Monday March 19, 2018 @01:47PM (#56284467)
      And yet, in most cases now they're orders of magnitude safer than the distracted meatbag texting away on their iPhone. Or the late-night drunk trying to make it home from the bar without getting caught. Yeah, even now I'd probably take my chances with the self-driving cars instead of humanity at the wheel, thanks.
      • Re:More to come (Score:5, Insightful)

        by Anonymous Coward on Monday March 19, 2018 @01:56PM (#56284565)

        There is no data to support your statement. Self driving cars haven't even started driving for real yet and Musk's marketing tweets don't count as a scientific study. Testing is not the same as real driving. At this time WHEN CONDITIONS GET BAD WE DON'T LET THE AI DRIVE. Let that sink in. If AI is so much better it should be able to outperform the worse the conditions get, not the other way around.

        Your statement is like saying you are a great basketball player but only during controlled ideal practice and you have never played a real game.

      • And yet, in most cases now they're orders of magnitude safer than the distracted meatbag texting away on their iPhone. Or the late-night drunk trying to make it home from the bar without getting caught.

        I'd love to understand your basis for saying that since they're not in widespread enough use to have generated any sort of meaningful statistics. And to OP's point, this may well have been one of the first situations where someone threw an autonomous vehicle a serious curveball. If so, that's at least an order of magnitude in the wrong direction.

      • Re:More to come (Score:5, Informative)

        by ichthus ( 72442 ) on Monday March 19, 2018 @02:01PM (#56284637) Homepage

        Yeah, even now I'd probably take my chances with the self-driving cars instead of humanity at the wheel, thanks.

        You have a very low opinion of your driving ability.

        • by Hadlock ( 143607 )

          In my experience, most people overestimate their driving ability by a factor of at least two. Very few humans drive when well rested, well fed and zero distractions and have perfect attention span.

        • Re:More to come (Score:5, Insightful)

          by barc0001 ( 173002 ) on Monday March 19, 2018 @02:51PM (#56285227)

          > You have a very low opinion of your driving ability.

          Nope, but I have a very low opinion of the driving ability of many people I see on the roads every time I drive somewhere. Self driving cars are probably already better than the lower 30% of licensed drivers out there and will only get better whereas that 30% will get worse as they age and let their bad habits get worse.

      • Re:More to come (Score:4, Insightful)

        by b0s0z0ku ( 752509 ) on Monday March 19, 2018 @02:03PM (#56284649)

        How many times a year does your computer freeze and need to be power-cycled, versus your brain doing the same. I hope they're using three redundant computers with separately written software, sort of like fly-by-wire aircraft do. The computers "vote" -- if one is out of whack from the other two, it's taken out of the loop.

        Even more important in cars since the separation distance between them and immovable objects tends to be measured in feet versus hundreds to thousands of feet.

    • All true...However it will take a couple of company busting lawsuits to stop the hype and inject some reality into the public.... and this will likely be the first.
    • by SuperKendall ( 25149 ) on Monday March 19, 2018 @01:51PM (#56284499)

      We're 50 years out from a working self-driving car.

      Thank you Mr Luddite. It's a shame that we currently live in a perfectly safe world where no pedestrians ever git hit and these darn self-driving cars come along and...

      Wait, what? Drivers hit pedestrians all the time? Especially so when they cross in the middle of the street at night in the rain?

      Remember, there WAS a human sitting behind the wheel. The fact that he didn't see here / could not react in time means she was (A) really hard to see, and (b) probably came in front of the car very suddenly.

      We are not 50 years from self-driving cars. We are *0* years from self-driving cars. They are being deployed today and the ramp-up will only continue, because even if they make mistakes it's still FEWER mistakes than people will make, on average.

      • by Joe_Dragon ( 2206452 ) on Monday March 19, 2018 @02:10PM (#56284745)

        a driver sitting ready to take over is not the same as one driving in manual mode

        • a driver sitting ready to take over is not the same as one driving in manual mode

          There are plenty of times drivers behind the wheel remove hands from controls to reach for something, or simply get sleepy, or are singing along to a song not paying attention... or maybe reactions are slowed because they are tired, or have had a bit to drink.

          The truth is a human driver out at 10pm at night in an empty downtown would not expect anyone either, and would almost certainly have hit the same women crossing in front

      • by ( 4475953 ) on Monday March 19, 2018 @02:20PM (#56284885)

        There is a huge difference between being killed or injured by a human driver and being killed or injured by a self-driving mechanism. In the first case, the human driver is either to be blamed or not. In the second case, you or your next of kin have to deal with a large corporation that is guaranteed to have top lawyers, and they will be constantly shifting the blame.

      • by jamesborr ( 876769 ) on Monday March 19, 2018 @02:45PM (#56285173)
        Having just had the joyful experience of driving over 200 miles on snow and ice encrusted highways with no pavement visible, no lane markers visible and enough active snow to obliterate whatever post delineators (snow) are out there (typically every 400-500 feet) -- especially at night, I personally would like to believe that the current driving AI's out there will or are mastering this type of driving -- I am just somewhat doubtful of the current "state of the art". This type of driving is fairly nuanced, particularly as the snow/ice surfaces rut up and changing lanes involve a very delicate hand -- and you can forget about any aggressive maneuvers (acceleration, braking or turning), even when the vehicle is not doing what you expect or need it to do and where reacting to problems too "assuredly" just results in even bigger problems. Unfortunately, this type of driving is fairly common in the northern parts of the country for 4-5 months a year. It'd be nice to "take the winter off", but that is not realistic and people have adapted to driving in these conditions (some better then others) -- and therefore, these autonomous AI's will either have to become just as proficient, or they just won't be adopted in all climates. Either starting (or more problematically during a trip) and having the AI "announce that due to some issue outside of it's control (no lane markers, snowed over sensors, insufficient traction, etc.)", it is unable to proceed will not be acceptable (nor be safe).
      • We're 50 years out from a working self-driving car.

        Thank you Mr Luddite. It's a shame that we currently live in a perfectly safe world where no pedestrians ever git hit and these darn self-driving cars come along and...

        Wait, what? Drivers hit pedestrians all the time? Especially so when they cross in the middle of the street at night in the rain?

        Remember, there WAS a human sitting behind the wheel. The fact that he didn't see here / could not react in time means she was (A) really hard to see, and (b) probably came in front of the car very suddenly.

        Or (c) probably avoided by an average driver, but the driver wasn't paying attention because the AI was in charge.

        We are not 50 years from self-driving cars. We are *0* years from self-driving cars. They are being deployed today and the ramp-up will only continue, because even if they make mistakes it's still FEWER mistakes than people will make, on average.

        Really? Do you have evidence for this?

        Do you have evidence that self-driving cars have lower accident rates when they drive under the same conditions?

        Given that human drivers are required in many scenarios, do you have evidence on how this breaks down, or what happens to total accident rates when the AI drives for 90% or the time and the human for 10%?

        Do self-driving cars mean people suddenly mu

    • Re: (Score:3, Interesting)

      by Anonymous Coward

      Weather was pretty clear in Tempe last night, no wind, dust or rain to speak of.
      That particular intersection is poorly lit and has a concert venue on one corner, an office building on another and then a very dark desert park on the other two.
      The fact that the human was not able to redirect the car either seems to indicate that they were caught by surprise as well

    • Re:More to come (Score:5, Informative)

      by rickb928 ( 945187 ) on Monday March 19, 2018 @01:59PM (#56284607) Homepage Journal

      the intersection of Mill ave and curry Rd in Tempe is not known to be commonly obscured by rain, snow, and at 10pm, not sunlight either.

      It is, however, a broad intersection, and the speed limit is either 40 or 50 MPH for both roads.

      It is also the site of a popular music venue, and a hip hop concert was booked then. Probably good attendance. They do serve alcoholic beverages and simple food.

      We don't know much, but I would expect this woman didn't use good judgement crossing this intersection, which requires a pedestrian to cross 6 lanes and bike lanes in every direction. It's not easy in the best of conditions, and if, God forbid, this woman was crossing without a walk sign, she was unwise. Hopefully the black boxes involved will share some info.

      • Re:More to come (Score:5, Insightful)

        by b0s0z0ku ( 752509 ) on Monday March 19, 2018 @02:08PM (#56284729)
        An argument for not building cities where cars come first at the expense of people who want to walk or cycle. Regardless of what's driving the cars.
  • by PeeAitchPee ( 712652 ) on Monday March 19, 2018 @01:44PM (#56284453)
    . . . to the existing legal system. So many have speculated what would happen when a self-driving car inevitably killed a 3rd party. Might as well get the process started so the litigation / legislation is resolved quicker and things move ahead . . .
    • by Bruce Perens ( 3872 ) <bruce@perens.com> on Monday March 19, 2018 @11:46PM (#56288943) Homepage Journal

      So many have speculated what would happen when a self-driving car inevitably killed a 3rd party.

      This isn't the case you're looking for. There was a driver behind the wheel, and he (or she) was responsible for the operation of the vehicle.

      I haven't heard of any good cases regarding autonomous mining trucks like CAT 794f, but those might come first.

    • by havana9 ( 101033 )
      There are a lot of people killed [dailymail.co.uk] by [nypost.com] elevators [dailymail.co.uk] every year.
      Normally chriminal charges are made to the company that installed the machine, the company that made the machine and the building owner.
  • by thegarbz ( 1787294 ) on Monday March 19, 2018 @01:48PM (#56284479)

    Let's all jump to conclusions. Don't disappoint me now. We should be at the root cause with all the information within the next 5 minutes.

  • Self-driving cars don't need to be perfect, just better than people.

    If self-driving cars rack up fewer pedestrian deaths per mile driven than human drivers, that's the critical metric.

    --PM

    • So what is the process for identifying and fixing the issue that caused this death? and how do you QA test that fix?
      In general, human drivers have a "main algorithm" when driving, but can easily switch to a "person in front of my car" subroutine. Arguably, in a high-pedestrian area, a human will load the "person in front of my car" subroutine into memory for quick access, anticipating unexpected humans.
      Human beings know that the assumption other humans will follow pedestrian laws is dumb. AI/self-driving ca

  • it can be blinding. It's called "Monsoon" weather. If you've never driven in it it's hard to explain. You can't see 6-8 feet in front of you. Like a white out but with water. Not sure if that's what happened here. I've been stuck driving in Monsoon rain a few times. I pull over as soon as I can and wait it out. It can be hard to do that with all the nut jobs trying to power through it.

    The woman was outside cross walks, so Uber will probably be in the right. Although IIRC you never have the right of way
  • by cogeek ( 2425448 ) on Monday March 19, 2018 @01:52PM (#56284523)
    So many missing details here. Was it raining causing limited visibility? Did she dart out between two cars right in front of the moving vehicle? Was she crossing and then doubled back? Did the human behind the wheel have time to try and react?

    The good thing about this being an autonomous vehicle is that there are likely cameras and sensors all around the vehicle that will be able to tell investigators exactly what happened.

    And while jaywalking is certainly not a capital offense, it's hard to argue that this would have happened if she'd been in a recognizable crosswalk with as many miles and hours as have been racked up by self-driving vehicles already.
    • Did she dart out between two cars right in front of the moving vehicle?

      No but close enough:
      Herzberg was "pushing a bicycle laden with plastic shopping bags," according to the Chronicle's Carolyn Said, when she "abruptly walked from a center median into a lane of traffic."

      After viewing video captured by the Uber vehicle, Moir concluded that “it’s very clear it would have been difficult to avoid this collision in any kind of mode (autonomous or human-driven) based on how she came from the shadows right into the roadway."

      https://arstechnica.com/cars/2... [arstechnica.com]

  • by SuperKendall ( 25149 ) on Monday March 19, 2018 @02:00PM (#56284627)

    While it's hard to say what happened yet, what I can say with confidence is that we should be able to figure out what happened far more easily than if any non-self driving car had hit her...

    Why? Because of the vast amount of sensor data collected by the car every second. We should be able to see exactly when she left the sidewalk, exactly where she went in the road, and exactly what led to the car not "seeing" her.

    Otherwise you'd MAYBE have some dash-cam footage and some super poor traffic camera footage.

  • by jfdavis668 ( 1414919 ) on Monday March 19, 2018 @02:13PM (#56284787)
    Now no one will be safe.
  • by foxalopex ( 522681 ) on Monday March 19, 2018 @02:17PM (#56284847)

    Despite the pedestrian jaywalking, Uber's had a LONG history of problems with their self-driving program. The worst google's self-driving program has gotten itself into is having cars crash into it because of confusion between right of way. Meanwhile Uber's managed to rollover one of their cars in a collision. At this point, I think Uber's rushing to have a successful IPO, Google is taking the time to do it right. So no thanks to Uber for giving self-driving cars a bad name...

  • by McGruber ( 1417641 ) on Monday March 19, 2018 @02:25PM (#56284927)

    The original reporting on ABC15 Self-driving Uber car hits, kills pedestrian in Tempe [abc15.com] actually includes a video that has the caption "Self-driving vehicle hits BICYCLIST". The video also shows a crumpled-up bicycle.

    Unfortunately, ABC15's text article says "a woman walking outside of the crosswalk was struck" and that is what the rest of the media is regurgitating as their own reporting.

    • by sinij ( 911942 ) on Monday March 19, 2018 @02:32PM (#56285005)
      Oh, that makes it all better. For a moment I thought that actual human was killed. Turns out it was just a bicyclist. I hit so many on may way to work that I have a windshield washer additive to help me clean the guts off the windshield.
    • by quenda ( 644621 )

      Verge says: "Early reports suggested that she may have been a bicyclist, but that was not the case."

      The road has a hard shoulder, and cycle lane at the nearby intersection.
      It sounds like the woman was jaywalking at night, pushing a loaded bike across a major road with no lights next to a park, instead of crossing at the nearby traffic light intersection. Does not make much sense. I'd wait for more info before blaming Uber on this one.

    • by thegarbz ( 1787294 ) on Tuesday March 20, 2018 @04:36AM (#56289527)

      A cyclist ceases being a cyclist when they are pushing their bike. Which is what was being done according to the police chief:

      Herzberg was "pushing a bicycle laden with plastic shopping bags," according to the Chronicle's Carolyn Said, when she "abruptly walked from a center median into a lane of traffic."

      After viewing video captured by the Uber vehicle, Moir concluded that “it’s very clear it would have been difficult to avoid this collision in any kind of mode (autonomous or human-driven) based on how she came from the shadows right into the roadway."

      https://arstechnica.com/cars/2... [arstechnica.com]

  • by Applehu Akbar ( 2968043 ) on Monday March 19, 2018 @03:25PM (#56285601)

    So Phoenix, as the Uber test city, has registered its first fatality. With no information made public other than it involved a jaywalking pedestrian, we have 250 posts predicting the entire future of the automated car industry. And illegal aliens, for some reason.

    I can't wait to see what the all-wise multitude will say once we actually know what happened.

  • Not surprised (Score:4, Insightful)

    by hdyoung ( 5182939 ) on Monday March 19, 2018 @04:34PM (#56286265)
    Google self-driving cars run literally millions of miles and the worst accident they get into is one of their cars getting rear-ended by somebody else. Uber gets into the game, and 3 months later they've killed someone. Can't say I'm surprised. Google is generally a responsible company. Uber uses a "break things, move fast, skirt the laws and let someone else pick up the wreckage" business model. Expect quite a bit more of this. I'm not opposed to rapid development of new tech like this. Sometimes, accidents will happen. 100% safety isn't a physical possibility. It's just that nobody should be surprised when outfits like Uber rack up an impressive body count.
  • Oh no: facts (Score:5, Informative)

    by Sinical ( 14215 ) on Monday March 19, 2018 @05:39PM (#56286923)

    She was hit here:

    https://www.google.com/maps/@3... [google.com]

    I know this because I looked at

    https://www.reuters.com/articl... [reuters.com]

    and I know the location intimately. The speed limit here is 40. The road, Mill Avenue, going northbound is two lanes plus it is adding turn lanes to go west and east. There is a bike lane. The road has just gone over a bridge (man-made lake) and under a freeway bridge (202) -- there are no off- or on-ramps at this location. There is a parking lot under the bridge for the concert venue (SW corner: visible in the Reuter's image) plus there's a public park/beach on the north side of the lake.

    As

    https://tech.slashdot.org/comm... [slashdot.org]

    states, there was no rain.

    http://alert.fcd.maricopa.gov/... [maricopa.gov]

    I haven't seen the crumpled bicycle photo, but we JUST started a bunch of "share bike" schemes in the Phoenix metro area (well, Phoenix proper has had one for while -- Tempe/Scottsdale ones are more recent): Limebike is the main one, I think (we have some that have "Ono" on them, as well). So if the bike is yellow or yellow/green, it was probably one of those. Tempe is hugely bike friendly for a US city because it is both (a) the site of ASU (b) progressive.

    The southbound lanes are 2 wide at this point, so this lady was riding a bike across ~5 lanes of traffic plus a BIG (mostly paved) median. There's a shortcut trail just RIGHT there to go east, so maybe she was aiming for that.

    A sad situation for sure. I see the Uber and Waymo vehicles all the time, so there's no lack of miles in and around that area.

Adding features does not necessarily increase functionality -- it just makes the manuals thicker.

Working...