Uber halts self-driving tests after pedestrian killed in Arizona

Started by tradephoric, March 19, 2018, 01:57:16 PM

Previous topic - Next topic

NoGoodNamesAvailable

Speed is also a nontrivial factor, even within a range of 3-5 mph above the speed limit. Your chance of dying as a pedestrian after being struck at 35 mph is 31%–compare that to 39% at 38 mph or 45% at 40 mph (reports of the vehicle's speed are conflicting). Considering the victim was hospitalized after being hit, 3-5 mph could have meant the difference between life and death.


tradephoric


Bruce

It's pretty obvious now that the pedestrian did not "dart" into traffic. She was already 2/3rds across and would be clearly visible to a responsible driver. And the driver could have stopped had the roadway been at a lower speed limit or given proper safety treatments (engineered for slower speeds, more lights added, more crosswalks).

Uber really did rush into this. If you read their statements during the Waymo lawsuit a few months ago, it's clear that they were trying to race their competitors and beat them. Highly irresponsible from them (but not unexpected of any car company...they don't care about safety) and from the government of Arizona for ceding public roadways as a test course.

kalvado

Quote from: tradephoric on March 21, 2018, 07:02:15 PM
Interior and exterior dash footage released of the fatal crash:
On one hand, pedestrian is not doing anyone a favor by walking across dark road in dark clothes. I am not sure what would be the outcome if I was driving.
On the other hand, this is exactly the situation where electronic driver should be able to perform much  better than a meaty one in avoiding an accident.

DaBigE

Quote from: Bruce on March 21, 2018, 07:54:38 PM
It's pretty obvious now that the pedestrian did not "dart" into traffic. She was already 2/3rds across and would be clearly visible to a responsible driver. And the driver could have stopped had the roadway been at a lower speed limit or given proper safety treatments (engineered for slower speeds, more lights added, more crosswalks).

Uber really did rush into this. If you read their statements during the Waymo lawsuit a few months ago, it's clear that they were trying to race their competitors and beat them. Highly irresponsible from them (but not unexpected of any car company...they don't care about safety) and from the government of Arizona for ceding public roadways as a test course.

Generalize much? And at some point, these vehicles have to be used in the real world. Would you be this critical of airplane development too?

You can't even see her until the vehicle is 50-ft away from her (assuming Tempe uses a standard skip/dash ratio). The SUV wouldn't be able to stop in that short of a distance even if they were driving less than the speed limit.

There are two things that clearly would have helped her in this situation: crossing at an illuminated crosswalk or wearing something with retroreflective elements to it. Looks like her bike didn't have the required reflectors on the wheels either.

Does anyone know if this was a hybrid SUV? It seems odd how long it takes the victim to look at the vehicle... did she even hear the SUV coming?
"We gotta find this road, it's like Bob's road!" - Rabbit, Twister

davewiecking

I feel sorry for the victim's family, who now has this woman's accident visible over the internet. However, she was clearly jaywalking at the time. She was pushing the bicycle, not riding it; I'm not aware of any requirement for pushed devices to have reflectors. (Do baby strollers have reflectors?)

The Goog tells me that Uber uses XC90's, which come with a hybrid option-front wheels gas; rear wheels electric (although I can't find out quickly if it has "fake external engine noise"). Interestingly, the XC90 also comes with Pilot Assist standard, which is a semi-autonomous driving (low-speed traffic jams as well as general highway driving; https://www.volvocars.com/us/cars/new-models/xc90). It also has Standard City Safety auto-braking technology, which detects imminent collisions and automatically brakes. Seems like the "stop before you hit something" technology was developed by Volvo, and Uber added the "get in the right lane and turn at the intersection" technology.

QuoteVolvo Cars' engineers have worked closely together with engineers from Uber to develop the XC90 premium SUVs that are to be supplied to Uber. The base vehicles incorporate all necessary safety, redundancy and core autonomous driving technologies that are required for Uber to add its own self-driving technology.
(https://www.media.volvocars.com/global/en-gb/media/pressreleases/216738/volvo-cars-to-supply-tens-of-thousands-of-autonomous-drive-compatible-cars-to-uber)

But I keep coming back to what I wrote 2 days ago. Why are there sidewalks in the middle of the median if they weren't to be used as such (and were signed to not be used)? In many places where authorities have recognized a problem middle-of-the-block crossing, they put up a fence in the median, because signs just plain don't work, especially in a poorly lit area at night.

I think the analogy with early airflight testing is a bad one. Although there may have been fatalities on the ground, those choosing to make early plane flights were well aware of the risks, and made the choice to fly anyway. I'm not sure many people moved to Tempe, AZ because they wanted to be guinea pigs for self-driving cars.

slorydn1

I believe I would have seen the pedestrian sooner from the drivers seat than it appears in the video just because our eyes do better than regular video cameras in low light situations, they just do. Even with that, I did see that a set of (headlights?) in the oncoming lane way in the distance get dim from left to right as an object crossed in front of it, and then the shoe of the pedestrian became visible in the same lane as the striking vehicle just as the clock changed from :02-:03. Saying that would be unfair to anyone driving this vehicle because I was a) on a second pass through the video and b) was expecting someone to suddenly appear in front of the car in a place they didn't belong. Notice the lack of any signs, markings, smoke signals, anything showing that this was a legal crosswalk.

I now can see why the car didn't pick up the conflict until it was too late, because she wasn't occupying the lane the car was in until the last instant. Are these things even capable of picking up and resolving low mass, low speed potential conflicts before they actually move into the path of the vehicle-something people have been doing since they first grabbed a steering wheel?

In any event, she didn't do herself any favors, notice that when she came into view she was looking down at the ground and away from the direction traffic would be coming? I am sure the car would have been visible to her long before she would have been visible to even an attentive driver.
Please Note: All posts represent my personal opinions and do not represent those of any governmental agency, non-governmental agency, quasi-governmental agency or wanna be governmental agency

Counties: Counties Visited

Bruce

Quote from: DaBigE on March 21, 2018, 08:45:19 PM
Generalize much? And at some point, these vehicles have to be used in the real world. Would you be this critical of airplane development too?

You can't even see her until the vehicle is 50-ft away from her (assuming Tempe uses a standard skip/dash ratio). The SUV wouldn't be able to stop in that short of a distance even if they were driving less than the speed limit.

There are two things that clearly would have helped her in this situation: crossing at an illuminated crosswalk or wearing something with retroreflective elements to it. Looks like her bike didn't have the required reflectors on the wheels either.

Does anyone know if this was a hybrid SUV? It seems odd how long it takes the victim to look at the vehicle... did she even hear the SUV coming?

There was a time where U.S. automakers tried to stifle new safety mechanisms and the unsafe nature of their vehicles, all because it would hurt sales.



Would you wear a reflective coat all day, every day? It's hard to buy one for every size while being comfortable. Heck, even a brightly covered coat is very hard to find, since everyone wears dark colors!

DaBigE

Quote from: davewiecking on March 21, 2018, 11:12:09 PM
I feel sorry for the victim's family, who now has this woman's accident visible over the internet. However, she was clearly jaywalking at the time. She was pushing the bicycle, not riding it; I'm not aware of any requirement for pushed devices to have reflectors. (Do baby strollers have reflectors?)

Quote from: §1512.16   Requirements for reflectors
There shall be retroreflective tire sidewalls or, alternatively, reflectors mounted on the spokes of each wheel, or, for non-caliper rim brake bicycles, retroreflective wheel rims. The center of spoke-mounted reflectors shall be within 76 mm (3.0 in.) of the inside of the rim. Side reflective devices shall be visible on each side of the wheel.
Source

If you regularly operate a baby stroller in the street, I would assume it would need reflectors as well. Crossing a street likely wouldn't count as "operating in the street" in the same manner a motor vehicle would.
"We gotta find this road, it's like Bob's road!" - Rabbit, Twister

DaBigE

Quote from: Bruce on March 21, 2018, 11:34:48 PM
Quote from: DaBigE on March 21, 2018, 08:45:19 PM
Generalize much? And at some point, these vehicles have to be used in the real world. Would you be this critical of airplane development too?

You can't even see her until the vehicle is 50-ft away from her (assuming Tempe uses a standard skip/dash ratio). The SUV wouldn't be able to stop in that short of a distance even if they were driving less than the speed limit.

There are two things that clearly would have helped her in this situation: crossing at an illuminated crosswalk or wearing something with retroreflective elements to it. Looks like her bike didn't have the required reflectors on the wheels either.

Does anyone know if this was a hybrid SUV? It seems odd how long it takes the victim to look at the vehicle... did she even hear the SUV coming?

There was a time where U.S. automakers tried to stifle new safety mechanisms and the unsafe nature of their vehicles, all because it would hurt sales.



Yeah, until automakers found they sell more vehicles the safer they are. Why do you think crash ratings get inserted into seemingly every car ad?

Quote from: Bruce on March 21, 2018, 11:34:48 PM
Would you wear a reflective coat all day, every day? It's hard to buy one for every size while being comfortable. Heck, even a brightly covered coat is very hard to find, since everyone wears dark colors!

Ever heard of a safety vest? You can even buy them at WallyWorld; large enough to fit over just about any coat.
"We gotta find this road, it's like Bob's road!" - Rabbit, Twister

TheArkansasRoadgeek

Quote from: Bruce on March 19, 2018, 11:37:21 PM
Also, the Uber was speeding (38 in a 35). Road diets would make it much harder for it to speed, and the limits can be reduced to a safer speed like 25 mph for arterials.
Wow three over! Come on! Most states have a 5 over lenience. 


iPhone
Well, that's just like your opinion man...

davewiecking

Quote from: §1512.16   Requirements for reflectors
There shall be retroreflective tire sidewalls or, alternatively, reflectors mounted on the spokes of each wheel, or, for non-caliper rim brake bicycles, retroreflective wheel rims. The center of spoke-mounted reflectors shall be within 76 mm (3.0 in.) of the inside of the rim. Side reflective devices shall be visible on each side of the wheel.
Source

Thanks for linking the source. For now, I'm going to respond with: Title 16, Chapter 8 (of which §1512 is a part) covers the Consumer Product Safety Commission, which relates to the condition of devices at the time of sale. I don't believe the CPSC relates to traffic laws. By your logic, if I have to replace a broken spoke and have to temporarily remove the reflector in order to do so, I'm in violation and the CPSC should site me for possessing such a thing, even if I'm sitting on my front porch.

DaBigE

Quote from: davewiecking on March 22, 2018, 12:19:45 AM
Quote from: §1512.16   Requirements for reflectors
There shall be retroreflective tire sidewalls or, alternatively, reflectors mounted on the spokes of each wheel, or, for non-caliper rim brake bicycles, retroreflective wheel rims. The center of spoke-mounted reflectors shall be within 76 mm (3.0 in.) of the inside of the rim. Side reflective devices shall be visible on each side of the wheel.
Source

Thanks for linking the source. For now, I'm going to respond with: Title 16, Chapter 8 (of which §1512 is a part) covers the Consumer Product Safety Commission, which relates to the condition of devices at the time of sale. I don't believe the CPSC relates to traffic laws. By your logic, if I have to replace a broken spoke and have to temporarily remove the reflector in order to do so, I'm in violation and the CPSC should site me for possessing such a thing, even if I'm sitting on my front porch.

By my logic? Ahh no. You're taking what I said a couple steps in the wrong direction. How is sitting on your porch the same thing as operating it on a public street? Just like it's one thing for your car to have a broken reflector or light while it's on your property. It's a completely different situation if you're driving said car down the street. The latter can get you pulled over and ticketed for an equipment violation.

Further, if you have a broken spoke, with the way most wheel reflectors are designed, you can easily relocate and reattach the reflector to a different location on the same wheel.
"We gotta find this road, it's like Bob's road!" - Rabbit, Twister

jeffandnicole

Quote from: Bruce on March 21, 2018, 07:54:38 PM
It's pretty obvious now that the pedestrian did not "dart" into traffic. She was already 2/3rds across and would be clearly visible to a responsible driver. And the driver could have stopped had the roadway been at a lower speed limit or given proper safety treatments (engineered for slower speeds, more lights added, more crosswalks).

Really?  At what point did you see the person? 

As for everything else you said: Nonsense.  What sense would it be to engineer the road for lower speeds?  There's nothing on either side of the road.  Why did the pedestrian decide to cross at this point?  There's no reason for the road to be given a lower speed limit and crosswalks be added where there's absolutely nothing around.  In fact, I think she's on a bridge or just off the bridge at this point.  As for lights on the road - again, there's no reason to light up every road everwhere.  It would be an incredibly inefficient use of lighting, and would dramatically contribute to light pollution.

It's unfortunate the driver was looking down, because it easily switches the focus off the car that was truly hidden until the last half-second. 

I'm pretty sure that people that want lower speed limits on every road will be equally unhappy once they realize that their buses and other means of transportation take much longer as well...because they have to go slower speed also.

signalman

I saw the dashcam footage this morning on the news.  I honestly did not see the pedestrian until the last second and I doubt very much that I would have avoided hitting her had I been driving.  While it's unfortunate that the pedestrian was killed, I believe that it was her inattentiveness and irresponsibility that led to her demise.

SP Cook

While most people hate trial lawyers, and there is a lot there to hate with their w***e witnesses spewing junk science and their obnoxious commercials, they do serve a purpose in our society.  At the end of the say lawsuit losses eventually change bad behaviors by corporations. 

Think of this.  This crew actually was willing to TEST an unproven technology on city streets around unconsenting people.  And now they have KILLED someone.

Hopefully everyone involved will be reduced to poverty and this dead end pipedream technology can be ridiculed for the load of crap it is.


kalvado

Pedestrian really did a lot to get hit; but there is still one big question:
WHY electronics didn't see her???  Looks like the car doesn't even try to brake within that last second.. And that is with sensor suit which is far superior to human eye...
Computer relies on visible light camera information only? Then ban such design from the road RIGHT NOW!
Lidar picked her up, but object wasn't recognized as dangerous? DOn't believe so.
Lidar was inop (e.g. dirt on glass)? Well, maybe. That should affect the operating standards quite a bit.
Black clothing was also too black in infrared, so lidar didn't pick her up? Plausible, but very disturbing...

kalvado

Quote from: SP Cook on March 22, 2018, 11:16:16 AM
While most people hate trial lawyers, and there is a lot there to hate with their w***e witnesses spewing junk science and their obnoxious commercials, they do serve a purpose in our society.  At the end of the say lawsuit losses eventually change bad behaviors by corporations. 

Think of this.  This crew actually was willing to TEST an unproven technology on city streets around unconsenting people.  And now they have KILLED someone.

Hopefully everyone involved will be reduced to poverty and this dead end pipedream technology can be ridiculed for the load of crap it is.

SO how do you prove technology without testing it in real world? Apparently someone did testing on closed range, and assumed thing is safe enough for testing in a wild.
I hope traffic laws violated by pedestrian would allow Uber to get out of this case with minimal losses. Amount similar to what would be paid in case of human driver accident - and even that is too much.

jeffandnicole

Quote from: SP Cook on March 22, 2018, 11:16:16 AM
While most people hate trial lawyers, and there is a lot there to hate with their w***e witnesses spewing junk science and their obnoxious commercials, they do serve a purpose in our society.  At the end of the say lawsuit losses eventually change bad behaviors by corporations. 

Think of this.  This crew actually was willing to TEST an unproven technology on city streets around unconsenting people.  And now they have KILLED someone.

Hopefully everyone involved will be reduced to poverty and this dead end pipedream technology can be ridiculed for the load of crap it is.



Imagine if someone tried inventing an automobile today.  The very first crash that occurs where someone would be killed would be the end of the auto, correct?

We have 40,000 people dying in auto wrecks every year.  And yet that is perfectly acceptable.  Amazingly, in this one crash, many people cite that if the human was paying attention, the crash could've been avoided.  And yet, we still have 40,000 auto deaths in which humans were at the wheel, in control of the vehicle.  Every year.

SP Cook

Quote from: kalvado on March 22, 2018, 11:22:03 AM


SO how do you prove technology without testing it in real world? Apparently someone did testing on closed range, and assumed thing is safe enough for testing in a wild.

And they were wrong.  And now they have killed someone.  Someone who had not consented to be a part of their little test.  People that get things wrong get sued.  People that test bad ideas on unconsenting subjects should get jailed. 

The way you test new technology is in the lab and proving grounds.  Of course setting up a proving ground for this complex a technology would cost billions.  So they just tested on whoever.  And now someone is dead. 

So, you would be fine if Kellogg's made a new cereal.  Maybe its poison, maybe not.  Why bother feeding it to some lab rats.  You just dig in.  And, BTW, we are not even asking your permission, just randomly inserting the new formula among all the regular stuff.  Interesting.


DaBigE

Quote from: SP Cook on March 22, 2018, 02:15:36 PM
Quote from: kalvado on March 22, 2018, 11:22:03 AM


SO how do you prove technology without testing it in real world? Apparently someone did testing on closed range, and assumed thing is safe enough for testing in a wild.

And they were wrong.  And now they have killed someone.  Someone who had not consented to be a part of their little test.  People that get things wrong get sued.  People that test bad ideas on unconsenting subjects should get jailed. 

The way you test new technology is in the lab and proving grounds.  Of course setting up a proving ground for this complex a technology would cost billions.  So they just tested on whoever.  And now someone is dead. 

So, you would be fine if Kellogg's made a new cereal.  Maybe its poison, maybe not.  Why bother feeding it to some lab rats.  You just dig in.  And, BTW, we are not even asking your permission, just randomly inserting the new formula among all the regular stuff.  Interesting.

You're assuming the real-world testing is the only testing that occurs. Of course there is going to be computer simulation/lab testing and closed-course tracks, but at some point, yes, the real-world will have to be used. The real-world is a never ending test, since there is no way you can ever account for every possible scenario.
"We gotta find this road, it's like Bob's road!" - Rabbit, Twister

SP Cook

Obviously they did some lab testing.  Not enough, because they just killed somebody. 

I get a kick out the pipedreamers who point out how many person driven cars have had accidents.  Umm, yeah, but this dead end technology has already killed several, with 0.001% of the mileage, and most of that in cherry picked environments and on cars with non-real world levels of vehicle upkeep.  And yet, PER MILE DRIVEN, this dead end technology kills 10000s for every one a human does.  It is just too dangerous to allow further. 

But even if that was not so, it remains totally unethical to test anything on non-volunteers.  I thought we learned that from the Tuskegee scandal. 

hotdogPi

Even if the average is 100 million miles (about the human average), there is about a 1% chance that it will happen within the first million miles. Are we sure this isn't the case?
Clinched, minus I-93 (I'm missing a few miles and my file is incorrect)

Traveled, plus US 13, 44, and 50, and several state routes

I will be in Burlington VT for the eclipse.

kalvado

Quote from: SP Cook on March 22, 2018, 02:15:36 PM
Quote from: kalvado on March 22, 2018, 11:22:03 AM


SO how do you prove technology without testing it in real world? Apparently someone did testing on closed range, and assumed thing is safe enough for testing in a wild.

And they were wrong.  And now they have killed someone.  Someone who had not consented to be a part of their little test.  People that get things wrong get sued.  People that test bad ideas on unconsenting subjects should get jailed. 

The way you test new technology is in the lab and proving grounds.  Of course setting up a proving ground for this complex a technology would cost billions.  So they just tested on whoever.  And now someone is dead. 

So, you would be fine if Kellogg's made a new cereal.  Maybe its poison, maybe not.  Why bother feeding it to some lab rats.  You just dig in.  And, BTW, we are not even asking your permission, just randomly inserting the new formula among all the regular stuff.  Interesting.

This is how you kill any innovation. Go to your cave and eat your raw wheat - flour mills are not proven to be safe enough, and stoves cause fires on a regular basis. And those studs in your house are treated with chemistry which may contain hell of a load of cancer-causing chemicals. They are tested to a certain level, but who said it is enough? Your home slowly kills you!
Anything new out there can have a glitch. Even Kellog cereal tested on lab rats can trigger allergy in someone - due to new processing technique of the grain, which causes problems to 1 person in a million.
Something more involved? Of course. Ever heard about car recalls? That is just that - lab tested technology not working in real life. Batteries burn, paints turn out to be toxic, and kids eat anything what fits in their mouth..
You may build a test city, hire 500,000 testers - but still not get someone wearing that specific brand and color of a jacket.

kalvado

Quote from: 1 on March 22, 2018, 03:05:04 PM
Even if the average is 100 million miles (about the human average), there is about a 1% chance that it will happen within the first million miles. Are we sure this isn't the case?
And you do expect MORE problems during testing than you get during regular use...



Opinions expressed here on belong solely to the poster and do not represent or reflect the opinions or beliefs of AARoads, its creators and/or associates.