Uber Car in Self-Driving Mode Kills Woman in Arizona

Photo:  Tim Templeton/Wikimedia
Photo: Tim Templeton/Wikimedia

An Uber car in self-driving mode has killed a woman in Tempe, Arizona. Details remain scarce and initial accounts vary as to whether the victim was walking or biking at the moment of impact. A human was behind the wheel but was not driving when the collision occurred, according to Uber.

This marks the first time a car in self-driving mode has killed someone outside the vehicle. But the detection of people walking or biking is a known weak point in the development of autonomous cars.

Arizona is one of the major testing grounds for self-driving cars, in part because the state has wide roads with relatively few pedestrians, which are easier for autonomous cars to navigate, though automotive companies are also testing the vehicles in more complex environments. Uber is suspending its autonomous vehicle testing in Tempe, Pittsburgh, Toronto, and San Francisco.

Now that someone has been killed by a self-driving car, questions that seemed like hypotheticals are all too real. Will Uber be held liable for this loss of life? Can these vehicles be safely tested in urban environments where people get around by walking and biking?

169 thoughts on Uber Car in Self-Driving Mode Kills Woman in Arizona

  1. I think she was cycling on the bike bath and was trying to get to the bike lane, which is in between the right turn lane and the straightaway right hand lane. The damage is to the front wheel bending right and the car damage is to the front passenger side.
    A lot of homeless people collect aluminum cans at night to recycle for money.
    The real issue is a mayor deciding to subject his citizens to being part of an experiment where the potential consequences are death or crippling. If you are going to use human beings as lab rats, which is not really moral to begin with, you could at least let them know they are lab rats.

  2. Unless I missed it. No one has mentioned the possibility that the car saw her, and I cant see how it couldnt have not seen her if she walked from the West, but was unable to determine what she was, because maybe the BAGS on her bike confused the image detection.

  3. That doesn’t make any sense: in what instance would a self driving car be programmed to run into an obstacle, even if it can’t determine what the obstacle is?

  4. If we set the bar at “as good as an average driver” then we’ll forever be stuck

    I really don’t see any reason that would be the case. Again, there’s precedent for the government mandating increased safety over time as technology allows for it. (And I would be pretty surprised if competition doesn’t outstrip that anyway; look what a big deal cars make of safety ratings in their advertising.)

    Why is there a rush to get AVs on the road?

    I wouldn’t call deploying something once it’s at a point where it’s better than humans at preventing serious injury and death “a rush”. I’d ask the opposite: if we could start saving lives now (assuming there are AVs at the point of being better than average human drivers—no argument at all that we should be demanding that bar), why would we wait and let more people die in the interim?

    There’s no reason at all that doing that has to conflict with also continuing to improve the technology and raising the mandated bar.

  5. Well for start, it might have caused a “long” delay. Research nvidia’s GPU driverless tech, on how it processes these images and the update frequency. Also this stuff is all being rushed. Which is the very reason Uber had to relocate to AZ. So who knows. Maybe the procedure is the operator is meant to stop the car when something happens that the car doesn’t understand? From previous reports that does seem to be the case in some instances

  6. An empty plastic shopping bag blowing across the street is the sort of thing that a car should just hit at speed rather than braking for. So is a cloud of dust.

  7. fancy math talk. back it up with hard evidence. we have to wait for these companies to get their act together. until then real lives are being beta tested on. these companies can test off our streets and then demonstrate why they should be allowed on. humans are smarter than these systems at the moment. get your tech geek glasses off for a second.

  8. Having watched the video it seems this woman died because of bicyclist entitlement syndrome. How else can you explain jaywalking a bike…at night…across a four lane highway…away from the street lights…in a black sweater…as a car speeds towards you at full speed?

  9. I know I said above I would give that other commenter the last word but I want to point out that the narrative of the woman jumping out of the shadows “in a flash” has been proven to be complete nonsense by the just-released video evidence. She was walking across the road and had already passed one lane and was just steps away from the shoulder when she was struck by the AV at regular speed.

    Yes, she’s wearing a dark jacket or sweater but the bike she was pushing was brightly-colored. It’s entirely reasonable that any driver–a computer-controlled car or a good old-fashioned human being–should have been expected to see her.

  10. Whether she was in a crosswalk is completely irrelevant. The story that she jumped out of “the shadows in a flash” is complete BS based on the video of the crash.

  11. you dont know the area. its common for people to cross their. funny how you analyse the person but not the car that is equipped with sensors that should function regardless of the amount of light present.

  12. video shows her walking the bike. yes i agree that being involutnarily a subject in a beta test is morally reprehensible and i dont care if these cars will some day be better than humans. right now they arent. get them off the streets. uber pulled them off the streets because they see how faulty they are. how about not letting them on in the first place until they actually are better?

  13. its clearly both. sensors designed to work independant of light didnt function properly. its well documented that they are poor at seeing cyclists and pedestrians. definitely the AV contributed to the problem not because of some philosophical premise that AV’s are all bad but because uber is new to the game and sucks so bad they decided to remove their cars from the streets.

  14. You’re absolutely right. I hadn’t seen the video yet when I commented. I had researched Waymo and didn’t realize how far off from its record Uber is. This feels like a massive failure for Uber and the lax approach to safety from Arizona regulators.

Leave a Reply

Your email address will not be published. Required fields are marked *