Video Released of Uber Self-Driving Car Accident

Having looked a few more times at the footage, to be an armchair driver, I believe a person paying attention would have noticed the pedestrian.

She was absolutely negligent in operating the vehicle, autonomous or not, and should be 100% liable and potentially face manslaughter charges.
She was not operating the vehicle. If she had been paying attention, she would have to react and take control of the vehicle. By then it would of been over.
I would be curious on the time it takes to take control of the vehicle. Stepping on the brakes would of done nothing but slow it a bit before hitting.
 
She was not operating the vehicle. If she had been paying attention, she would have to react and take control of the vehicle. By then it would of been over.
I would be curious on the time it takes to take control of the vehicle. Stepping on the brakes would of done nothing but slow it a bit before hitting.

This is full of crap, you can take control of a autonomous vehicle by taking the wheel (I've tried them). If the Uber system did not allow that then fair point, all others do.

She was not paying attention at all, there is physical evidence of her negligence, she is at fault.

Uber bares some fault as well, but she was the operator, her job was to monitor the vehicle, not play crazy cupcakes.
 
This is full of crap, you can take control of a autonomous vehicle by taking the wheel (I've tried them). If the Uber system did not allow that then fair point, all others do.

She was not paying attention at all, there is physical evidence of her negligence, she is at fault.

Uber bares some fault as well, but she was the operator, her job was to monitor the vehicle, not play crazy cupcakes.
Yes, and is it instant or does she have a reaction time? That is the key factor, even if paying attention.

Maybe some calculations on the factors would help. I bet they would say she would still have been hit.
 
Yes, and is it instant or does she have a reaction time? That is the key factor, even if paying attention.

Maybe some calculations on the factors would help. I bet they would say she would still have been hit.

I bet not, having seen enough video that person was visible from some distance. Again the negligence is proven in the video.
 
I bet not, having seen enough video that person was visible from some distance. Again the negligence is proven in the video.
Yes it is. The automated system didn't work well when a person runs in front of a vehicle at night and not in a crosswalk. ;)
 
A lot of people in this thread seem to think high-res, high-freq, sensors are off the shelf items. They are not. They are restricted and I would imagine they aren't going to allow Joe Blow to buy a self-driving car and harvest the parts needed to make an ad-hoc missile.

you mean highly restricted sensors like this? installed on a lot of cars now?

flir-pathfindir_100193552_m.jpg
 
The driver is a backup for the system. That is why they are there and it is MANDATED by law in most states. And if you actually WATCHED the video, you would have realized it was a woman who was doing nothing but playing with some gadget with her eyes off the road 1/2 the time.

The safety driver was pretty much doing exactly what she was suppose to. SDs are there for gross faults in the AV system not to micromanage the actual driving.
 
The safety driver was pretty much doing exactly what she was suppose to. SDs are there for gross faults in the AV system not to micromanage the actual driving.
I don't know about you, but striking a pedestrian sounds like a gross fault to me.
 
There was a strange area of total darkness covering all the left lane. When she first appeared in the visible spectrum, she was already well into the car's lane. I doubt that I would have been able to brake in time.

However, the visible spectrum shouldn't be the main sensor for the car at night. I wish that they showed the infrared view.


EDIT: This guy on Twitter has a ton of situational photos.
Pictures of a controlled intersection just 100 yards away. Pictures of a deceptive "beautiful brick-paved diagonal walking path" that has a sign saying it should never be used, and aerial views of the street. It looks like this was a perfect storm where absolutely everyone did everything wrong: the pedestrian, the car, the security driver, and the infrastructure.
That doesn't look like a walking path to me, it looks like a turnaround for emergency services.
 
the cars ladar should of seen it (it does not give any hoots about light conditions ) the main issue is the car did not brake at all when both the visual and ladar should of been visible to both (especially last 2 seconds)

the driver should of been looking forward (unsure on what will happen to him)
 
I think infrared technology should definately be there for cases like this. The Volvo would have had HID or LED projector headlights with a sharp beam cut off and no stray light.

A human would have hit her regardless, it was a combination of factors that lead to it. Personally what idiot pedestrian doesnt notice a car bearing down on them on a dark road with no other traffic.
 
Then you're an idiot who has no experience driving. It's a straight road, no obstructions, car traveling at 40 MPH. If the headlights were working properly, a human driver paying attention would have plenty of time to see the woman entering the road and react appropriately.

I live in Tempe. I know this environment. This accident shouldn't have happened, period.

Uber will fry for this and they deserve to. They are going to get hammered in court (or settle for 7+ figures) and their self-driving tech is going into the garbage can.

There's a good chance this is related to Uber's alleged theft of the self-driving technology they are using.
If they'd developed it themselves, they might actually understand it, but if they just stole it ... not so much.

It's also related to Uber's corporate culture of breaking the law and putting people at risk (like by hiring felons as drivers) in the name of corporate valuation.


I have over 35 years experience driving, and I can tell you, that you are a complete tool. The lights where working properly on the vehicle, as shown in the video, she came out of the dark shadows with only a couple seconds to react (actually only 1.5 seconds). There is no way in hell anyone could have not hit the person period.

Now, as my son and I where talking about this earlier today, one thing came to mind is the difference between driving on a freeway, and driving in town. There is a higher chance of someone walking out in front of you in town, as well as more things to monitor for. I suspect that there is a completely different set of protocols the self driving car uses when on the freeway vs in town. You may not realize it, but we as humans do the exact same thing. We shift our attention to farther up the road, because we are traveling at a much higher speed, and we are looking for hazards farther ahead to give us enough time to react and maneuver. We operate on a completely different set of protocols when driving on a freeway vs in town, just as i suspect the self driving car does. Even when driving in town, driving a lower rate of speed, it would have been difficult to miss the person coming out of the dark shadows with only 1.5 seconds to react, much less being able to avoid it on a highway.

Now, why is it that there is a 3 second rule when following another car on the highway? Because it takes that much time to react to something the car in front of you may do. Keep in mind, that is 3 seconds following another vehicle that is going the same speed and direction as you are.. NOT a NON moving object. (yes the pedestrian was moving, but her direction made her basically a non moving object because she was not going the same direction). Even the 3 second rule doesn't prevent accidents, it just gives a person time to react to something the car in front of you may do.

So, you are somehow saying that you could have reacted, swerved and/or stopped in 1.5 seconds or 85 feet? Keep in mind the average reaction time is 2.3 seconds when driving in the city and stopping for a red light , which is .8 second to late in this situation, at night, in the dark, on the highway. http://copradar.com/redlight/factors/index.html
 
Last edited:
  • Like
Reactions: Croak
like this
just some random youtube video (and Chinese at that , 4 years back ) showing the detection range of sensors = 80-150m

why would UBER's specs be any worse than this. It's just waiting for a fatal mistake to happen .



-----------------------------------------------------------------

and anybody tasked to authorize on the road testing would have at least reviewed that closed-circuit tests has proven capability of the car to detect and avoid collision like this (or deer collisions) . Plenty of test tracks around. (no lights, sudden obstacles / crossing obstacles / static obstacles = all very easy to simulate)
 
Last edited:
Would a non-self driving car strike the person in the video under these circumstances? I'm willing to bet yes since it was in the dark, they were in a 2 lane road with no traffic going at around 40 mph. The woman doesn't seem to be crossing anywhere she should be and wasn't wearing reflective clothing at least that can be seen in the video.

If you start thinking that autonomous vehicles should be programmed in every instance or be able to come to a complete stop/avoid hitting people no matter what, then you're asking too much. I'm willing to bet that the car isn't programmed to look out for people when on a highway.
IMO that person appeared out of nowhere, not sure a human would've done better.
 
I have over 35 years experience driving, and I can tell you, that you are a complete tool. The lights where working properly on the vehicle, as shown in the video, she came out of the dark shadows with only a couple seconds to react (actually only 1.5 seconds). There is no way in hell anyone could have not hit the person period.

Now, as my son and I where talking about this earlier today, one thing came to mind is the difference between driving on a freeway, and driving in town. There is a higher chance of someone walking out in front of you in town, as well as more things to monitor for. I suspect that there is a completely different set of protocols the self driving car uses when on the freeway vs in town. You may not realize it, but we as humans do the exact same thing. We shift our attention to farther up the road, because we are traveling at a much higher speed, and we are looking for hazards farther ahead to give us enough time to react and maneuver. We operate on a completely different set of protocols when driving on a freeway vs in town, just as i suspect the self driving car does. Even when driving in town, driving a lower rate of speed, it would have been difficult to miss the person coming out of the dark shadows with only 1.5 seconds to react, much less being able to avoid it on a highway.

Now, why is it that there is a 3 second rule when following another car on the highway? Because it takes that much time to react to something the car in front of you may do. Keep in mind, that is 3 seconds following another vehicle that is going the same speed and direction as you are.. NOT a NON moving object. (yes the pedestrian was moving, but her direction made her basically a non moving object because she was not going the same direction). Even the 3 second rule doesn't prevent accidents, it just gives a person time to react to something the car in front of you may do.

So, you are somehow saying that you could have reacted, swerved and/or stopped in 1.5 seconds or 85 feet? Keep in mind the average reaction time is 2.3 seconds when driving in the city and stopping for a red light , which is .8 second to late in this situation, at night, in the dark, on the highway. http://copradar.com/redlight/factors/index.html

except there were no dark shadows

Untitled.jpg
 
It was the bicycle. The pedestrian walking with a bicycle by its side made the system ignore it completely. So with uber's POS system if you walk with a shopping cart, a baby stroller, or are on a bicycle, a scooter, a tricycle.. bam! You are dead.
That video is bullshit, i have shot videos that look exactly like that, with crappy cameras.. reality is vastly different.. there was plenty of light around. Plenty.
 
Last edited:
Is it just me or does the woman in the car look more like a bloke with a mullet instead ?
 
Some of the responses in this thread make me cringe. Let's talk about failures in this situation, in order of priority.

1) The state failed.
Did the state do its due diligence before allowing autonomous cars on the road? Did it require the autonomous vehicle to pass a series of tests before deciding it was acceptable to allow them on the road? Did the state provide a protocol the autonomous vehicles are required to follow during the testing? Was the protocol adequate? Was the protocol followed?
I think the state bears some responsibility for allowing its citizens to be subjected to woefully inadequate systems which can kill someone. (As a note, some of the above questions I do not have the answer to and really would like to have those answered.)

2) The car failed.
The car failed to detect an object moving in a collision vector with it. Nothing else needs to be said here. This was a gross systems failure resulting in someones death.

3) The safety driver failed.
No doubt the safety driver was not paying attention to the road. Why didn't the states test protocols require the safety driver to take command of the car every X number of minutes in order to insure full participation of the safety driver? It is a well known fact humans will fail if suddenly thrust into a dangerous situation requiring a decision to be made with only seconds of exposure to the situation. Why did the test protocol allow the driver to have anything which would serve to distract them from the task at hand?

4) The pedestrian failed.
A homeless person, possibly incapable of adequately judging closure rates of oncoming vehicles, moved in front of a moving car.
If everything else was working the way it was supposed to have worked, this final failure would not have happened.

To blame the pedestrian for the total failure is, at best, ignorant, at worst it is woefuilly neglectful. If Uber/Arizona is lucky, they will have a load of people who think that on the jury.
 
Is it just me or does the woman in the car look more like a bloke with a mullet instead ?
Because he is Rafael Vasquez who did a 4 year turn in prison for attempted armed robbery and then decided to be a... whatever that is. Cant make any assumptions these days.
 
Person wears dark clothing at night, right after a street light. A person in the car's place would've struck her.

She's not even detected in the video frames until the car's almost right under the left street lamp.
 
Person wears dark clothing at night, right after a street light. A person in the car's place would've struck her.

She's not even detected in the video frames until the car's almost right under the left street lamp.

Scroll up 5 posts to see that section of road is far more well lit than the uber video presents.
 
Because he is Rafael Vasquez who did a 4 year turn in prison for attempted armed robbery and then decided to be a... whatever that is. Cant make any assumptions these days.

The women is a bloke ?

Dafuq
 
240 lines is not enough to make a decision off of. That's a crap image, especially when you are trying to determine course and heading with a computer.


So how many lines IS enough? I just grabbed a random FLIR image to demonstrate that the vehicle should have been able to detect that woman long before it did, there are sensors with much higher resolutions available right now, including 640 line 90 degree models.
 
Well perhaps the software did see the person and looked at it as either it could do a massive swerve which would result in the drivers death or that the best option was to hit the passenger. This was one of the big dilemmas with autonomous vehicles. If that person just stepped off the curb just before the car was there then the vehicle may have just made the choice.

If it was a human driver as the police chief stated he does not believe that a human would have been able to much either. A human most likely would have attempted to probably try and avoid perhaps but that we will never know.
 
Yes it is. The automated system didn't work well when a person runs in front of a vehicle at night and not in a crosswalk. ;)
What video were you watching? Because the one posted shows a walking victim, not a running one. And what does a crosswalk have to do with it, does Ubers sensors detect faded crosswalk paint better than it detects a bicycle and ambling human?
 
Last edited:
A driver paying attention might have been able to brake a bit but to say they could have completely avoided her is just you talking out of your ass. That dumb bitch bears the major responsibility here crossing the road in the dark like that where she shouldn't be crossing.
So sorry for your anger at the inadequacies of self-driving cars being publicly out-ed... I'm guessing this will considerably slow the development of "killer" cars. Does this mean you will have to drive your own car for the next 20 or 30 years? FYI: yes...
 
She was not operating the vehicle. If she had been paying attention, she would have to react and take control of the vehicle. By then it would of been over.
I would be curious on the time it takes to take control of the vehicle. Stepping on the brakes would of done nothing but slow it a bit before hitting.
Humans can also use the steering wheel. The combined effects of applying the brakes and swerving might have resulted in a glancing blow.
Then again the vehicles autonomous parts should have been able to do the same thing. Instead it plowed into her and it looks like the brakes were never actuated with zero steering input.
 
It was the bicycle. The pedestrian walking with a bicycle by its side made the system ignore it completely. So with uber's POS system if you walk with a shopping cart, a baby stroller, or are on a bicycle, a scooter, a tricycle.. bam! You are dead.
That video is bullshit, i have shot videos that look exactly like that, with crappy cameras.. reality is vastly different.. there was plenty of light around. Plenty.

I just watched the video too and I agree. Human eyes have decent latitude in the dark. And when I can't see well in the dark, that's what the brights are for. It's strange that not the driver but the car did not flash the brights in this case which leads me to believe it was not so dark. Considering the video came from the car's camera, how do we know it was not altered to make it look darker? We don't know and that extreme shadow is quite exaggerated. That said, it's a fail for an autonomous car to hit a PED in the middle of the street w/o traffic jaywalking or not. It just doesn't add up.

And from the pov of the jaywalker, I would venture to guess that given how empty the road is they did not expect to be hit and that they could see the car from distance.
 
Under those circumstance, a human driver would have hit her too.

Then you're effin blind. You can see someone in the road far enough away to not hit them... at night... at less than 40mph. Sorry.

Go to your volvo dealership and test drive their XC60 at night. That video is not an accurate representation as to what your eyes will see. This is a massive failure for uber, i hope they pay a steep penalty.
 
So how many lines IS enough? I just grabbed a random FLIR image to demonstrate that the vehicle should have been able to detect that woman long before it did, there are sensors with much higher resolutions available right now, including 640 line 90 degree models.

That's the 64,000 dollar question. It depends on how fast you are going, how far you want to see, and what kind of processing you are going to do with it. Sensor pitch and interpolation can mean missing perfectly horizontal or vertical edges, especially on small arrays. How fast those arrays can be reliably dumped matters too. None of which ultimately matters since the Uber system seems to only mount visible-light cameras.
 
I think the real issue here, is not so much Uber's software failed... I mean yeah that's tragic and all, but if Uber purposefully doctored that video that says loads more and that should have Uber engineers, CEOs, etc facing criminal charges.

That said, I still find it funny that all these article writers (who strangely are basically saying exactly what some are saying here) are not grasping the fact that an $800 phone might have a better camera and/or software for adjusting to low light conditions by making it looking brighter than say a cheap dashcam for the car, and I can get away with saying cheap because the dashcam doesn't in any way affect how the autonomous driving features work (no data is used from it) so why would you put a super high end camera in? Now maybe they purposefully put the crappiest dash cam they could just for instances like this. But either way you know Uber's lawyers are preventing the release of all the sensor data that the car actually saw.
 
so spending millions on r&d, cheaps out on dash cam or enhances the footage so you can see nothing. well it is uber the scummy company.
 
I think the real issue here, is not so much Uber's software failed... I mean yeah that's tragic and all, but if Uber purposefully doctored that video that says loads more and that should have Uber engineers, CEOs, etc facing criminal charges.

That said, I still find it funny that all these article writers (who strangely are basically saying exactly what some are saying here) are not grasping the fact that an $800 phone might have a better camera and/or software for adjusting to low light conditions by making it looking brighter than say a cheap dashcam for the car, and I can get away with saying cheap because the dashcam doesn't in any way affect how the autonomous driving features work (no data is used from it) so why would you put a super high end camera in? Now maybe they purposefully put the crappiest dash cam they could just for instances like this. But either way you know Uber's lawyers are preventing the release of all the sensor data that the car actually saw.

The higher the resolution, the higher the storage requirements especially for video. They probably had it set low & crappy so they could fit it on whatever the max SD card size that dashcam supports is. Why they didn't spring for a high-end model (for something as important as research review) is a pertinent question.
 
The driver is a backup for the system. That is why they are there and it is MANDATED by law in most states. And if you actually WATCHED the video, you would have realized it was a woman who was doing nothing but playing with some gadget with her eyes off the road 1/2 the time.


The driver is required by AZ law to move a malfunctioning vehicle off the roadway and shut it off. They are not a safety requirement for a situation like this. Right now, fully autonomous trucks are making deliveries all acrossed AZ with no safety drivers or humans at all in the vehicles. These vehicles are supposed to have redundant systems to perform the function of the Uber driver which is why they are not needed.

Governor Ducey just signed his Executive order on the issue a month or so ago and there is nothing in it about needing a safety driver to react in emergency situations like this, nothing about having the be prepared to assume control should the system fail.

Even then, what he did sign is far more than what was required previously.


EDITED:

I was mistaken about the Uber Trucks, there is an Uber employee in the seat to monitor the vehicle's systems, but they say he is not a driver and he is not there to drive the vehicle.
 
Last edited:
Leave the self-driving cars in the manufacture plant. Not looking forward to the future with these everywhere.
 
We can blame the lighting issue on NHTSA and the fact they haven't allowed Volvo (or any other manufacturer) to use their active high beams. Had they been in use, their wouldn't have been an unlit area to the left front of the car since the active high beam would light up that area until an oncoming car came and the beam adapted around the oncoming car. Instead the antiquated headlight laws still have vehicles dipping their beams down and to right so they don't dazzle oncoming traffic. That's no longer an issue with the advanced lighting systems implemented in vehicles but they still haven't been approved for use in the US.

It still doesn't excuse where this woman decided to cross the street, but it certainly could have illuminated her earlier.
 
Back
Top