Any details on headline - Arizona pedestrian is killed by Uber self-driving car

[zmarty]

I think this is the road where Tesla's autopilot would engage. How would it act in this situation, lets say if the driver was not paying attention to the road?
We did not have cases like this one that just happened with Uber happen with Tesla

Let's ask an even broader question: Would cars with basic standard safety features like AEB based on radar have attempted to stop in this case? And note I mean attempted, so at least break to lessen the impact. My guess is YES.

 

[buttershrimp]

Uber Video Shows the Kind of Crash Self-Driving Cars Are Made to Avoid

“I think the sensors on the vehicles should have seen the pedestrian well in advance,” says Steven Shladover, a UC Berkeley research engineer who has been studying automated systems for decades and watched the video. “If she had been moving erratically, it would have been difficult for the systems to predict where this person was going,” he says, but the video shows no evidence of that.

This video makes me angry, angry with Uber. This is a clear worst case scenario in my opinion. I am baffled by this. You are testing a fully autonomous vehicle on public roads. You are actually employing the person who is responsible for watching the road... and serving as backup. This scenario can't happen. It must not happen. Yes it could happen to anyone growing comfortable with the technology while supervising the drive, but the investment is so significant for Uber, if the driver actually had any record whatsoever, it is a massive failure for Uber. This is akin to finding out a Tesla employee killed a pedestrian IN TESTING of autopilot! This is a disaster for Uber and will set back autonomous car testing quite a bit. Poor woman.... and I feel for the driver too, can't imagine what any of them are going through.

 

[zmarty]

if the driver actually had any record whatsoever, it is a massive failure for Uber.

I was on board with you until this sentence. The record of the person should not matter.

We cannot blame the safety driver 100%. If the safety driver is to blame 100% then what blame does the Uber software have, 0%?

If the video is indeed representative of how the scene looked like (although I doubt that), I could probably not have stopped in time myself. But I expect the system to stop because it is using at least 3 types of sensors.

 

[McRat]

From the video I can't see how Lidar wouldn't have been able to see her crossing. I can also say that it appears the headlights are really horrible on that Volvo. And it appears to me that the "safety driver" was using her phone and only occasionally looking up at the road. (So it is no wonder that she was reported as having said that the first she knew of anything was when she heard the accident.)

That was 'driving faster than it is safe'. The headlights covered only 1 second of road. Should have had high beams on.
But the car sensors should have seen it as both cross traffic and as an object straight ahead.

Safety driver was not watching the road.

 

[dhanson865]

Holy crap, when I read about this the first day I gave Uber the benefit of the doubt, especially with the police saying it couldn't be avoided.

After watching that video I blame Uber and the driver behind the wheel that was looking down at a cell phone 80% or more of the time the video played.

I'd have no problems with convicting the person in that drivers seat for being at fault for the car hitting that pedestrian (though I'm not sure what the appropriate charge and punishment should be). I'd have no problems with Uber being fined on top of that.

I'd abosulely decide in favor of the deceased victims family if it were a civil lawsuit case.

 

[Cheburashka]

Wow, yeah that pedo should have been very visible .I see pedos and animals in ambient night light all the time. Straight road, no fog, slow speed, should have been avoidable.

The driver is a moron and Ubers self driving sucks.

 

[Cheburashka]

That was 'driving faster than it is safe'. The headlights covered only 1 second of road. Should have had high beams on.
But the car sensors should have seen it as both cross traffic and as an object straight ahead.

Safety driver was not watching the road.

I think headlights were fine. Ubers camera or video is just terrible quality, probably on purpose.

 

[brkaus]

@dhanson865 agree - the comments under the video say "driver looking at console" but I highly doubt it. Looks more like phone. I'd hang them just for that (and calling them an engineer).

I do think the sharp & low headlight cutoff plays a role. I had a rental last week that was similar - I hated it.

And of course, the person had no business walking across the street there when they could clearly see a car coming.

So there is blame to be shared.

 

[glhs272]

So I just watched the video for the first time just now. Some thoughts. Wow, what a horrible tragedy. I feel great sympathy for that person who lost their life. Watching that person appear from the black right in front of the car is eerily similar to an incident in my past where a I hit a deer. An oncoming car just past me and my low beams were on and suddenly as the car past, BAAM there is a deer. No time to react. Deer dead, car wrecked. In this case there was no oncoming car but there were street lights contributing to some of that effect. It would definitely make it harder to identify an object that is just past the street lights in the darkness. To me it looks like there is some blame to go around. For one, the self driving car should do better than us humans at seeing these objects. This is where radar should have seen through the darkness to recognize the object moving into the path of the car, beyond the headlights beams. Also, it was obvious that the human backup was distracted (most likely a phone) and was no longer being a backup to the system. Lastly, it appears that the person with the bike did not equip their bike with adequate reflectors on their wheels. If they had, the driver might have seen that bike long before it appeared out of the darkness. The wheel reflectors will reflect the faint light long before the main beam of the headlights illuminates the bike. I attached a picture showing there appeared to be no wheel reflectors on the bike.

 

[dhanson865]

@dhanson865 agree - the comments under the video say "driver looking at console" but I highly doubt it. Looks more like phone. I'd hang them just for that (and calling them an engineer).

I do think the sharp & low headlight cutoff plays a role. I had a rental last week that was similar - I hated it.

And of course, the person had no business walking across the street there when they could clearly see a car coming.

So there is blame to be shared.

I think the pedestrians punishment was too severe. Or put another way time served, no need to punish them more.

I can't say the pedestrian was without blame, just that I don't think the pedestrian was more to blame than the other parties. So I focus my attention on the ones remaining to learn from their mistakes and be punished since the price paid by the pedestrian can't be refunded.

 

[SMAlset]

I think it a fair bet that either (I) the video that's on the internet was taken by a dashcam that's entirely separate from the cameras used by the self-driving system or (ii) the video did come from the self-driving camera, but (probably to reduce the storage space used over a long period of testing) contains vastly less frames and resolution than is available to the self-driving software from the camera.

This video of the road was from the car's mounted dashcam. Police had mentioned there was a front and rear camera and they looked at those. I'm sure the police do not have Uber's camera video yet. I know from our own Blackvue DR650S that our eyes pick up way more than the camera does after reviewing some of our videos. Heck I have these Sunbeam blue light nightlight/safety lights plugged in across the way from our bed and in pitch black with the small blue LED lights I can see into the other room just fine. Really believe if the "safety" driver was paying attention to the road for purposes of taking over if necessary she would have seen the pedestrian. Given that the pedestrian was as far in the lane as she was when you see her in the dashcam video, I think she could have swerved to the left to avoid her and missed her if they had been paying attention. Clearly the car, while we can't hear any audible sounds from the dashcam video, didn't alert the safety driver as she didn't look up until, as she said, she heard the impact.

Not sure what the camera was that was focused on the safety driver was. Could it have also been a third party IR camera that taxi's use to monitor their driver and passengers? It seems to be recording from behind the rear view mirror.

 

[Nikxice]

Just watched the interior camera video focused on the driver. Thirteen seconds long. Head down for eleven, eyeballs outside for just two. That habit would eventually bite any driver. The future path of this vehicle can only be observed a second or two ahead, requiring the safety driver to have 100% of her attention outside. I won't judge whether that would have saved the life of the pedestrian, however shame on Uber for not periodically auditing tape and screening for poor driving skills.

 

[Tam]

...our eyes pick up way more than the camera does...

Exactly!

Human eyes pick up images in the dark much better than what the video above shows.

There were 2 street lights above that shone much dimmer than the car's headlights did so the video automatically darkened the whole view so it wouldn't be overexposed.

Not so with our human eyes. Our eyes can pick up the image in low light that's shown by those 2 dimmed streetlights above.

Without Uber's equipments, an attentive driver could see way much further all the way to the dimmed streetlights before the video can pick it up.

I've experienced with quite a few of jay walkers in the dark in my life exactly just like this case. My headlights would shine very limited but my eyes could pick up moving images far way more beyond my car's headlights. My saving grace was that I was lucky that I paid attention at those times.

The above scenario is very much not fatal if the driver was attentive and if the driver could only see what the headlights saw, there's still plenty of time to apply the brake.

1) Legally, without crash avoidance technology and video footage, a driver would most likely get away scot-free because of the nature of jaywalker even in good daylight.

2) However, the video shows the driver's distraction, so getting away scot-free is now very much remote.

3) In addition to driver's distraction, the public was expecting some performance from all those collision avoidance technology with its great redundancy but the system didn't react and resulted in a collision which would complicate the otherwise surety of getting away with fatal accident scot-free.

 

[barjohn]

If you go to Galveston, TX during Mardi Gras or the Lone Star Rally, you will see scenarios like this repeat several times each year and these are not autonomous vehicles. People try and cross the main beach road called the Seawall at night between street lights and they just don't realize how invisible they are. There are at least two occasions I came close to hitting pedestrians/bicyclists there and every year 3 or 4 get killed. They try and run across the street and often they have been drinking and misjudge the speed of the car coming toward them. They also think because they can see the car that the driver can see them. Bicyclists wearing dark clothes, no lights or reflectors and virtually invisible until you are about to hit them. You normally have on low beams due to the heavy salt air mist or on coming traffic. What I saw in that video was that she was invisible until she was in the cars headlights and due to the vehicles speed and her proximity there was no time to do anything. It takes about ½ second to recognize what you are seeing and hit the brakes. Think about the following scenario. You are following behind another vehicle. You are at a safe distance (at least 2-3 sec separation) but it is a dark night. You have your low beams on. Suddenly the car in front swerves, you are startled and just as you are trying to figure out what is happening you see a big object in the road the car is swerving to avoid hitting. You wind up hitting it. Why, because you didn't have enough time to understand and process what you were seeing and take evasive action or hit the brakes. Similar scenarios have happened to most of us one time or another and we hope to hell we didn't damage or break something.

 

[barjohn]

The street lights can make it worse when someone is in the cutoff area as she was. The sharp contrast between brightly lighted area and the dark area makes it hard for our eyes to see into the dark areas. Sure there are a few people with really great vision and dynamic range that might be able to see into the dark area but they are the exception rather than the rule.

I just stepped the video through the frames and she went from not being visible to being hit in 1 sec. so no human would have been able to respond that quickly. Maybe the computer should have seen her with LIDAR or radar sooner and reacted to stop the vehicle. Also she crossed the path so quickly that it appears she might have been running and thought the car was further away or moving slower than it actually was.

 

[Tam]

...no human would have been able to respond that quickly...

I am sorry but it's no INATTENTIVE human could respond timely.

She did respond with her mouth wide open but her hands were no where near the steering wheel and they seemed to be moving away more from the steering wheel and I assume her foot was no where on guard to apply the brake either.

If she complied to the protocol and be attentive at the steering wheel and brake, I don't think this collision would be fatal.

She needs to be trained how to react in emergency: It's ok to open her mouth but her hands and foot need to be in correct positions even if it's the last second or even if it's too late!

I am not trying to blame her because I have been inattentive many many times in my life but I have been lucky that I was attentive during jaywalker encounters.

I know my luck will not be guaranteed forever so I hope Tesla technology will be able to take care of this scenario soon and I believe Waymo did just that but not Uber!

 

[barjohn]

If there had been more time maybe, but she becomes visible at the start of second 03 in the frame and is struck at second 04 in the frame. Very hard to comprehend and react in 1 second. (young athletes excepted)

 

[Evoforce]

I am sorry but it's no INATTENTIVE human could respond timely.

She did respond with her mouth wide open but her hands were no where near the steering wheel and they seemed to be moving away more from the steering wheel and I assume her foot was no where on guard to apply the brake either.

If she complied to the protocol and be attentive at the steering wheel and brake, I don't think this collision would be fatal.

She needs to be trained how to react in emergency: It's ok to open her mouth but her hands and foot need to be in correct positions even if it's the last second or even if it's too late!

I am not trying to blame her because I have been inattentive many many times in my life but I have been lucky that I was attentive during jaywalker encounters.

I know my luck will not be guaranteed forever so I hope Tesla technology will be able to take care of this scenario soon and I believe Waymo did just that but not Uber!

From how I saw the video synced, she didn't open her mouth until impact.

 

[zmarty]

Waymo Engineer: “Watching this made me sick. Irresponsible driver + a flawed system. If you can't detect a pedestrian standing in the middle of the road, you shouldn't have self driving cars on the road.”

https://twitter.com/vahidk/status/976627819450007552?s=21

 

[zmarty]

It certainly looks bad for Uber | Brad Ideas

"I have seen one report -- just a rumour from somebody who spoke to an un-named insider, that the LIDAR was off in order to test operations using just camera and radar.

[...]

The road is empty of other cars. Here are the big issues:

1. On this empty road, the LIDAR is very capable of detecting her. If it was operating, there is no way that it did not detect her 3 to 4 seconds before the impact, if not before. She would have come into range just over 5 seconds before impact.
2. On the dash-cam style video, we only see her 1.5 seconds before impact. However, the human eye and quality cameras have a much better dynamic range than this video, and should have also been able to see her even before 5 seconds. From just the dash-cam video, no human could brake in time with just 1.5 seconds warning. The best humans react in just under a second, many take 1.5 to 2.5 seconds.
3. The human safety driver did not see her because she was not looking at the road. She seems to spend most of the time before the accident looking down to her right, in a style that suggests looking at a phone.
4. While a basic radar which filters out objects which are not moving towards the car would not necessarily see her, a more advanced radar also should have detected her and her bicycle (though triggered no braking) as soon as she entered the lane to the left, probably 4 seconds before impact at least. Braking could trigger 2 seconds before, in theory enough time.)"