I suspect that they will find that Uber wasn't liable. The article from the SF Chronicle has more details:
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.”
There is an envelope around any car where physics dictates that there's nothing any driver can do, human or robot, to stop a car from hitting something that is suddenly inserted into its path. To attempt avoidance in such a situation might lead to a far worse accident - or example, sharply swerving in an attempt to avoid a pedestrian might cause the vehicle to lose control and strike other vehicles or pedestrians. There's no magic software patch that's going to fix that.
The chief of the Tempe Police has told the San Francisco Chronicle that Uber is likely not responsible for the Sunday evening crash that killed 49-year-old pedestrian Elaine Herzberg.
[...]
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."
Moir added that "it is dangerous to cross roadways in the evening hour when well-illuminated, managed crosswalks are available."
[...]
At a Monday afternoon press conference, Tempe Sgt. Ronald Elcock said that there were no signs that the vehicle slowed down before striking Herzberg.
[...] even if Uber is not legally at fault, critics may ask whether the vehicle could have done more to anticipate the possibility that the woman might step in front of the Uber vehicle and take appropriate defensive measures—perhaps slowing down or changing lanes to give her a wider berth.
Dumb Person Gets Hit By Car. Film at 11.
Dumb Person Gets Hit By Car. Film at 11.
I would replace "dumb" with "careless". We are talking someone who just died, here.
Chumpy_McChump wrote:Dumb Person Gets Hit By Car. Film at 11.
I would replace "dumb" with "careless". We are talking someone who just died, here.
"dumb" is probably a needless insult, but it is a woman who essentially j-walked a busy road at night in the dark from a concealed median. i think you have to call that "reckless" at a minimum.
I'm having a hard time getting the facts straight. I would like to see the accident site on street view to understand it better.
I just don't understand how a person standing in a median can be invisible to a normal car driving down the road with the lights on. A human driver would be watching and anticipating a non-moving pedestrian who was out of place in a median that isn't meant for pedestrian crossings. Which is definitely something an automated car would not do.
even if Uber is not legally at fault, critics may ask whether the vehicle could have done more to anticipate the possibility that the woman might step in front of the Uber vehicle and take appropriate defensive measures—perhaps slowing down or changing lanes to give her a wider berth.
That could work once almost all cars are autonomous and can talk to each other, until then if an automated car did this everyone would complain about it slowing down for no reason or swerving into their lane.
I just don't understand how a person standing in a median can be invisible to a normal car driving down the road with the lights on. A human driver would be watching and anticipating a non-moving pedestrian who was out of place in a median that isn't meant for pedestrian crossings. Which is definitely something an automated car would not do.
Posted about this somewhere in the distant past of this thread. There will be some teething problems while people learn what kind of behaviors are unrecognizable by automated cars that would be obvious to a human.
Some of those behaviors will be patched in to be recognizable, and some will be things that people learn not to do any more as greater percentages of cars on the road become automated. Back before the automotive revolution, streets were places to wander and pedestrians roved everywhere. When these really fast steel boxes that could kill you started appearing in numbers, crosswalks sprang up and pedestrians learned how to watch for speeding objects at greater distances before entering the street. Same thing will happen here.
I'm having a hard time getting the facts straight. I would like to see the accident site on street view to understand it better.
I just don't understand how a person standing in a median can be invisible to a normal car driving down the road with the lights on. A human driver would be watching and anticipating a non-moving pedestrian who was out of place in a median that isn't meant for pedestrian crossings. Which is definitely something an automated car would not do.
?
It's like you've never driven on a poorly lit street at night with pedestrians wearing dark colors.
I actually have. I also have a masters degree in transportation engineering and am a certified road safety auditor.
Which is why I'm asking for more information on the site of the crash. Where I'm from accidents in 35 mph areas rarely involve deaths.
(PS. On a side note I'm also really curious how a self-driving car is able to drive 38 mph in a 35 mph area. That is some pretty brilliant programming right there)
(PS. On a side note I'm also really curious how a self-driving car is able to drive 38 mph in a 35 mph area. That is some pretty brilliant programming right there)
That is actually the least worysome part to me. I can't speak for everywhere, but in my area a vehicle actually going 35 on a non-residential 35mph road causes a pretty significant traffic disruption. Having the automated vehicle going closer to 5 over would cause much less of a problem than ones going right at the speed limit, at least until the majority of cars are automated.
None of the self driving cars treat speed limits as sacred, because that's a really dangerous way to drive (in the US), instead they will drive the speed of traffic unless conditions mean that the safe speed is lower than that.
I have no idea what there various criteria for "traffic" are, and when they make the cutoff of "go the actual safe speed based on conditions, or go the speed limit". For example if the last car on the street with them turns off do they slow down to the speed limit right away, wait a couple minutes and then gradually slow down over another couple minutes?
One of the secondary things I'm actually excited about with self-driving cars is a fleet of vehicles autonomously moving around the streets picking up data like that. There are a couple intersections near me where I strong suspect a bush limits the visibility enough to be unsafe or even illegal, having a big company actually be able to empirically verify that sort of thing and make complaints to change speed limits or improve visibility would be nice.
I actually have. I also have a masters degree in transportation engineering and am a certified road safety auditor.
Which is why I'm asking for more information on the site of the crash. Where I'm from accidents in 35 mph areas rarely involve deaths.
(PS. On a side note I'm also really curious how a self-driving car is able to drive 38 mph in a 35 mph area. That is some pretty brilliant programming right there)
Several articles point to just south of the Mill Ave/Curry Rd intersection, with the Uber car traveling northbound at the time. Do with it what you will.
Well it looks like there is a pathway through the median... but at the point the path meets the road there is this:
Some of those behaviors will be patched in to be recognizable, and some will be things that people learn not to do any more as greater percentages of cars on the road become automated. Back before the automotive revolution, streets were places to wander and pedestrians roved everywhere. When these really fast steel boxes that could kill you started appearing in numbers, crosswalks sprang up and pedestrians learned how to watch for speeding objects at greater distances before entering the street. Same thing will happen here.
It's worth noting here that jaywalking was a term created by the auto industry to put blame onto pedestrians using the streets as they always had and not on the two-ton death machines suddenly cruising them. We've ceded a lot of space, particularly in cities, to cars. One of my worries in a future of connected and automated vehicles is that takeover being completed. These cars will always have to respond to human behavior.
In response to comments about the speed:
Well it looks like there is a pathway through the median... but at the point the path meets the road there is this:
![https://i.imgur.com/IOXcJDD.jpg[img]](https://i.imgur.com/IOXcJDD.jpg[img]){kind=link}
Thanks for the link. As far as I can see, there are lights along this road which contradicts the news reports. But perhaps they are turned off in this part of this city during evenings/nights? Or maybe I'm just looking the wrong place...
Creating pathways for pedestrians (irregardless of signage) and planting trees/bushes that reduce visibility to and from the pathways is definitely a poor choice in road design under all circumstances. These are classic factors leading to accidents.
Having looked at it closer I'm not completely convinced that a driver had no chance of reacting. Even a slight decrease in speed has a dramatic effect on how serious injuries are when you are in the 35 mph area (look at the graph in the post above). It will be interesting to follow this when/if more facts are made available to the public.
My image of the sensors on these cars may be inaccurate, but I was under the impression that lighting wasn't as important to see obstructions like this, rather than street and sign markings. I would have thought that the radar would not have picked her up, the lidar would have picked her up, and thus the car would know there was a "soft" thing (aka person, dog, etc) at that location. Unless by "shadow" they mean the shadow of a pillar or something in between the pedestrian and the car.
I actually work about a mile from this location and drive through it regularly.
The median here has a lot of vegetation that can block your view of pedestrians. I wouldn't call it dangerous, but the responsibility is definitely on the pedestrian's end. Phoenix in general is one of the most dangerous cities in the country for pedestrians, we have a lot of jaywalkers and the like and cars speed regularly. Also, the area has one of the highest concentrations of homeless in the entire Valley, so that opens up a whole range of possible contributions to the accident.
Zoomed in satellite view of the area south of that intersection. There's a lot of vegetation, and I can see at least two places where a pedestrian could easily be hidden from a car before they stepped out. Worse, you can see from the paths in the image that the area is frequently used by pedestrians, and apparently that's encouraged by the footpaths that are built into the median even though they are marked with warning signs for pedestrians not to use them (which I'm sure is a bit of story in itself).
I believe those bushes would have masked the Lidar that would have seen the woman. (And of course seeing her and being prepared for her to step in front of the car without notice are two different things).
Tempe police have released the video. (Video stops immediately prior to impact, but may be difficult to watch)
Based on the video, the pedestrian had already crossed one lane of the road and was moving into the second lane when she was struck. The video makes it fairly obvious that a human driver would not have seen her in time, but it seems she should have been easily detectable by the car's lidar which, if I understand the technology correctly, should not have been affected by the dark. It'll be interesting to see what Uber's investigation reveals.
It's always going to be this way, if the news had to dramatically headline every human driver caused injury or death in the same fashion every time they occurred people would be taking to their governments demanding human drivers be banned and self driving mandated, alas unlikely to occur in reality - whatever the facts may be, easier to fear monger :/
I'm wondering if human eyes actually would have been better in the dark than an ordinary camera. I'm not so certain an ordinary camera is a good metric for determining if something is in fact visible when it is not daylight. Any thoughts on this?
At the same time the culprit in this road design is clearly poor lighting with too long a distance between lights creating dark spots in between. This is not possible where I live but perhaps there aren't uniform codes for this in the US. It's also unfortunate that the lights were not placed where pedestrians are crossing the road (illegally). The road designers obviously knew where this was as they had made walkable pavement in the median. Seems like an obvious oversight.
Even if some Human eyes COULD see better in that situation many do not. I and many of us probably know plenty of older folks not allowed to drive after certain hours because their night vision, even with city lighting simply isn't good enough anymore. Furthermore at these point I doubt many autonomous cars have 'normal cameras', given they can easily have sensory input with a 360 sphere of night vision+infravision+sonar+radar+speed of electricity transit and reaction (versus our very slow in comparison speed of chemistry)
In response to comments about the speed:
I won't comment on the detail of this incident, as I think this is one for the appointed investigators rather than casual speculators.
But I will note that speed limits in West London are generally being lowered from 30mph to 20mph for precisely the reasons set out in ActualDragon's graphic.
Initially, these limits were set on streets with schools (a UK advertising campaign from a few years ago referred explicitly to the chances of a child surviving impacts at different speed). However, they have now been extend borough-wide (at least in Ealing and Hammersmith & Fulham). I expect these limits to be applied to urban roads to be nation-wide before too long... And I cant' see a good reason why they shouldn't.
30 kph in the Philippines is mandated in the traffic codes in any area frequented by pedestrians. Not that anyone reads the Traffic Code, but hey, it's there.
I'm wondering if human eyes actually would have been better in the dark than an ordinary camera. I'm not so certain an ordinary camera is a good metric for determining if something is in fact visible when it is not daylight. Any thoughts on this?
If it was actually dark, yes - human eyes are reasonably well adapted to the dark. However, driving on a road with periodic street lighting and using headlights destroys any semblance of human night vision.
The video definitely brings up the question of why the sensors didn't see the pedestrian - she clearly wasn't stepping out in front of the vehicle from cover, but was in the road and should have been sensor visible.
I agree with Aetius, if only from an anecdotal perspective. My personal experience on a road with other cars, especially with oncoming traffic, my night vision is adjusted to the level of the streetlamps and oncoming headlights so unless it is a completely unlit country road any space not directly lit might as well be invisible. I suspect this effect is why walking a road at night in dark clothing is more dangerous than in bright clothing which is more dangerous than in retro-reflective clothing.
I can't watch the video because Twitter won't load, but wasn't she pushing a bicycle? I wonder if the metal of the bicycle had an impact on how the car evaluated what this thing in the roadway was.
Kaos, she is pushing a bike, but is walking on the side closer to the car, so the bike is not between her and the car. She is wearing dark clothes and there is a shadow demarcation line right where she is walking, so on the video anyway she is not visible until the car is already there.
In my opinion the video shows definitively a case of the pedestrian being at fault because there is no way she looked properly for cars and there is no crosswalk. Of course it also shows the 'safety driver' was repeatedly doing something down in her lap, probably playing with her phone. Assuming the video is accurate of the view, I don't think a person would have been able to avoid the accident either.
Pages