Daisuke Wakabayashi, New York Times:
Arizona officials saw opportunity when Uber and other companies began testing driverless cars a few years ago. Promising to keep oversight light, they invited the companies to test their robotic vehicles on the state’s roads.
Then on Sunday night, an autonomous car operated by Uber — and with an emergency backup driver behind the wheel — struck and killed a woman on a street in Tempe, Ariz. It was believed to be the first pedestrian death associated with self-driving technology. The company quickly suspended testing in Tempe as well as in Pittsburgh, San Francisco and Toronto.
Ugh.
A preliminary investigation showed that the vehicle was moving around 40 miles per hour when it struck Ms. Herzberg, who was walking with her bicycle on the street. He said it did not appear as though the car had slowed down before impact and that the Uber safety driver had shown no signs of impairment. The weather was clear and dry.
How was this vehicle allowed on the road? Something as big as a person with a bicycle, and slow (the woman was walking her bike) and the autonomous camera system did not see it? Tragic. And more so, because it never should have been allowed to happen.
UPDATE: Hat tip to Loop regular Drew Leavitt for this San Francisco Chronicle article, titled Exclusive: Tempe police chief says early probe shows no fault by Uber. From the article:
Pushing a bicycle laden with plastic shopping bags, a woman abruptly walked from a center median into a lane of traffic and was struck by a self-driving Uber operating in autonomous mode.
“The driver said it was like a flash, the person walked out in front of them,” said Sylvia Moir, police chief in Tempe, Ariz., the location for the first pedestrian fatality involving a self-driving car. “His first alert to the collision was the sound of the collision.”
Sounds like no driver could have avoided this accident. But:
Traveling at 38 mph in a 35 mph zone on Sunday night, the Uber self-driving car made no attempt to brake, according to the Police Department’s preliminary investigation.
The car was exceeding the speed limit (barely, true, but the autonomous programming allows this?) and it made no attempt to brake. Which tells me that the system did not detect the pedestrian. Seems to me an autonomous driving system should be able to react in milliseconds. Was this a blind spot in the system?
All this said, I do get that no system is perfect. And that an autonomous system has the chance to be much, much better than the human driver it replaces. But when something like this happens, it makes me feel like a flaw has been exposed, and an opportunity to improve is ours for the taking.
There is just not enough information about this accident yet to pass judgement. Was she in the actual road? It was night, was there any lighting in the area? Was she wearing anything light or reflective? Why didn’t the human safety driver see her either?
I know I’ve been out on the road at night and come up on people walking/riding on the side of the road that I couldn’t see until I was right on top of them.
Reactions like “How was this vehicle allowed on the road?” before the facts are in are problematic. Autonomous vehicles will make the roads much, much safer. Humans are terrible drivers as far too many are distracted, selfish, and reckless. Overreacting to any accidents that may occur during testing and development of this technology will cause more harm than good.
There will be many more people killed by autonomous vehicles, all progress requires sacrifice. Nobody wants it to happen, but unfortunately, humans act in ways that can’t be simulated and the learnings from this will help the systems get better.
The problem, however, is that Uber probably won’t share what improvements they make to their system better to other companies working on the same things.
I think there needs to be collaboration, at a minimum, on building better safety systems that everyone can learn and improve on before more people are killed.
Our world since 1981: privatization of profits; socialization of losses. I suspect the victim’s family will shortly find out that Arizona has a “business friendly” tort code which prevents them from winning a lawsuit against the perpetrator for any significant amount. Can’t have business deterred from sacrificing others to its gain.
I agree that businesses have far too much freedom to do what they want to feed the value of their stock.
I also think that autonomous vehicles can provide overall safety improvements, which is why I think we’d be better off if companies were mandated to cooperate and share knowledge in building autonomous systems.
I doubt that would happen, which is why I say they should do it at a minimum related to safety systems. Remove safety systems from being a profit/marketing advantage.
She was hit on a 6 lane road “And as soon as she walked into the lane of traffic, she was struck by the vehicle.” Any time a pedestrian walks into a traffic lane, there is a good chance for trouble. I don’t think that we should blame the victim, but stepping out in front of ANY car is a bad idea.
“Reactions like “How was this vehicle allowed on the road?” before the facts are in are problematic.”
Agreed.
“Autonomous vehicles will make the roads much, much safer. Humans are terrible drivers as far too many are distracted, selfish, and reckless.”
Agreed #2. We have to remember that humans kill *thousands* of other humans yearly with cars. While one terrible incident like this is one too many, we need to keep it in perspective.
Could a hacked self-driving vehicle mow down pedestrians, intentionally targeting as many as possible? Or disabled en masse to cause disruption. Or during a mass evacuation, all exits being blocked by rogue autonomous vehicles? Not a reason to not develop the technology, but certainly things that should be considered now, not after it happens.
The catch is that the road from here to there is probably paved with a lot of these deaths.
I think it needs to be travelled because humans are so very bad at driving, but it’s not going to be drama free.
Not to get into larger arguments about AVs, but I’m a cyclist in Pittsburgh, so I’ve been living with these things for over a year now. Honestly, they’re not noticeably different in behavior from other cars—including doing dumb stuff sometimes.
But the thing I think about most is the stories I’ve read of people riding in them who describe how they regularly hit the brakes, fairly hard, for non-events. And the thing is, they’re big, auto transmission SUVs. And when you take your foot off the gas in such a vehicle, it doesn’t really slow—the thing where auto trans are always sending some power to the wheels means that you have to hit the brakes to slow.
So what this means is that Ubers don’t seem to have a setting between maintaining speed and hitting the brakes. I drive a manual, so when I lift my foot, I immediately drop some speed, especially at city speeds. If I see something I don’t like—distracted pedestrian, loose toddler, car backing out a little quick—I lift my foot and bleed speed without committing to a stop or near-stop. And I just wonder whether Uber gets this. Nothing I’ve seen or read suggests that they do.
Isn’t that exactly what we’ve been told is the advantage of driverless vehicles? That a vehicle with this technology would “see” and respond to externalities humans generally have more trouble with?
Testing will resume of course, and eventually these vehicles will ready for roadways, but this is a somber reminder that real people are on the receiving end of this experiment.
What’s arguably different here is the arena in which this technology is being tested. Generally, potentially dangerous technologies are tested in areas roped off from the public.
Surprisingly, to me anyway, self driving vehicles have moved from private property to public roadways with seemingly unprecedented speed.
Its a sacrifice when the volunteer Uber attendant gets killed when the car drives off the edge of a cliff. When the car hits an unsuspecting bystander, its just a person getting killed by a car.
Uber should make these vehicles easy to see, especially in low light. A bright white, red, or yellow would be better than a dark grey.
It’s not magic. An autonomous vehicle won’t hit, or not see, a pedestrian because it’s distracted by a cell phone or the kids in the back seat. It’s always paying attention to the road, which all by itself is a HUGE step up over humans. However, if a person is on the road in such a way that they can barely be seen, if at all, then autonomous or not it’s a recipe for disaster. (To be clear, I do not know that was the case here.)
We’re on the same page: that’s the advantage and the ultimate goal of FAVs.
Definite case can be made for out-driving your headlights. The failure of the the sensor system to miss the pedestrian on such an open road is Very troubling…I remember a Cadillac commercial that showed their sensors picking up a deer well out of view of the headlight beam…don’t remember if there was a disclaimer on whether it was staged. Video of the incident is attached.
http://www.thetruthaboutcars.com/2018/03/uber-tempe-video/#more-1617988