noisebeam

At least until AV are commonplace.

CliffordK

Shouldn't have been on the road meaning that she shouldn't have used the beautifully paved "not a sidewalk" sidewalk?

Or, shouldn't have been on the road meaning that she attempted to cross a 2 or 3 lane road with inadequate time before the next vehicle came? How far away was the Uber car when she committed? Were its lights visible? It could well have traveled 1/4 mile in the time it took her to cross the first two lanes.

I was actually surprised at how much difference to visibility crossing under a street lamp would have made.

A car should avoid running over things. And, in many cases, one can safely avoid running over dogs, cats, deer, and other wildlife too. Stopping for a deer doesn't mean totaling one's car. Not stopping for a deer could lead to significant front damage.

Recently I was on a 2-lane road. I think a cat was in my lane. Car in incoming lane. I was happy to see that both myself and the car reacted to the possibility that I could scare the car to go back left across the car's path, and the car slowed well ahead of the possible critical moment.

KD5NRH

If "should" and "is" could be counted on to coincide on any sort of regular basis, there wouldn't be any need for AV sensor arrays. You could just use a simple navigation algorithm and match it to traffic light timings.

njkayaker

More self delusion.

I-Like-To-Bike

Good write-up today from Bloomberg https://www.bloomberg.com/news/artic...-arizona-crash
includes information about limitations of analyzing the accident by only looking at the dash cam video and also takes issue with the "exclusive" comments from the Tempe police chief


Extract:
"This is similar to the average reaction time for a driver. That means that, if the video correctly reflects visible conditions, an alert driver may have at least attempted to swerve or brake," Smith said.

The comments contrast with those made by the Tempe police chief, who told multiple media outlets that the pedestrian moved suddenly in front of the car and the crash didn’t seem preventable after reviewing footage of the collision.

"It’s very clear it would have been difficult to avoid this collision in any kind of mode,” Sylvia Moir, the police chief in Tempe, Arizona, told the San Francisco Chronicle.

Moore, the forensic engineer at Wexco, said dashcam videos tend to understate what humans drivers can see. While the pedestrian appears from the shadows in the video, a human driver may have had a better view if they’d been watching, he said.

Sean Alexander, of Crash Analysis & Reconstruction LLC, concurred. "Video makes everything in the light pattern brighter and everything out of the beam darker. A human eye sees it much clearer," he said.

Alexander also agreed with Moore’s analysis that a human driver could have avoided hitting Herzberg. "During the time the vehicle should have been braking, the pedestrian would have had additional time and would have cleared without the vehicle actually having to stop," Alexander said.

Tempe police released a statement Tuesday saying that "fault has not been determined in this case" and that a decision on criminal charges would be left to county prosecutors once the investigation is complete.

I-Like-To-Bike

The apologists always use that excuse

Every bicyclist who has the misfortune of being in front of a moving motor vehicle, no matter where they were located on the road; they just seem to come out of nowhere.

Daniel4

Although I disagree with a couple of statements, I do agree that the back up needed to be alert at all times.

In all systems involving public safety, the backup or backups need to be fully operational.

In Ontario in the May 2000, the government had cut budgets so much, water inspection failed and seven people died. The back-up was not in place to catch the front line inspectors falsifying the water quality tests.

Ironically, human driven cars have no backups.

McBTC

I'd have thought radar would have worked better than only seeing as far ahead as a driver with low beams on so, perhaps an equipment malfunction there; but, the pedestrian surely could see those headlights.. I'm thinking a jury might think... yikes, that could have been me behind the wheel and that guy came out of nowhere... wonder what the BAC was of the pedestrian.

AlmostTrick

True. Then again, there aren't dozens of companies attempting to take over the roads with robots that throw metal objects into bike wheels, so the potential threat is nowhere near the same.

Right. Had Elaine been hit by the left side of the car, the above suspicion wouldn't have been as likely, but she wasn't. She was near the right side before being hit. The fact that she finally looked to her right at the last second also suggests that an additional second may have given her enough time to survive by dodging.

But what about all the peds and cyclists who recover completely from lower speed collisions? I'll take my chances there, thank you... no matter which side of the front bumper I was on.

SHBR

It would appear that the video was released to place blame squarely on Elaine, however this may not go as planned.

If the public has any sense at all.

I do know from my own videos that I can see MUCH farther ahead than any dash or action camera, day or night.

I often am turning or braking for hazards well before they appear on video.

I also agree that there were three points of failure in this incident, the most significant failure is the backup driver, followed by the AV, and Elaine's choice to cross without checking traffic.

That said, there are animals, children, and even people who do this ALL the time, I deal with it on a daily basis.

I've never hit any animals, children, cyclists, or pedestrians, (lots of close calls though) and I hope I never do.

Maelochs

So ... after all this, the takeaway remains, that it is surprising that the AV did not spot her. All these pages of sturm und drang to get back to the beginning. BF .......

scott967

I know I was expecting to see a ped jump off the curb, but that doesn't seem to be the case. I hope there is recording of the sensor data on the Uber. With an apparent clear FOV, should have been within sensor distance to at least attempt to stop or avoid, not just plow through.

scott s.
.

Maelochs

I too wondered if the lady was under the influence.

In any case, in terms of liability for the accident, the car/driver can be absolved ... but in terms of the effectiveness of the AV operating system, it seems Uber has a substandard product.

CliffordK

Hard to have true statistics from a single incident.

However, Uber has apparently had about 2 million miles of autonomous vehicle tests, and a single fatal accident, or about 50 fatalities per 100 million miles.

On average, it is about 1.18 per hundred million miles (all fatalities by vehicle).

https://www.washingtonpost.com/opini...ce8_story.html

It could be a random occurrence, but it may well be that humans are better at avoiding running over other humans.

AlmostTrick

What we don't know is how many times a fatality was avoided over those millions of miles when a human driver intervened and took control. The robot cars haven't exactly earned their records on their own.

Ald1 

Just watched the video. Very disturbing. One, the driver was not paying attention, looking down until the last second. Two, the car appears to NEVER take evasive actions or brake until after the impact, even though the victim walked across an entire lane BEFORE walking in front of the car with maybe 50-60 feet in front of the car. As pointed out, I thought these cars were suppose to be better than us with infra-red sensors etc? We seem to think everything with a computer chip in it is "new and Improved." Our grid is in danger from cyber attacks precisely because of this while China and other "under developed" contries are not as much as theirs are mainly mechanical.
Just because we can does not mean we should.......

jefnvk

Is it a good writeup, or is it simply a media outlet with an agenda to smear Uber? I mean, if you wanted me to believe the conspiracy that that the Tempe PD and SF Chronicle was working to protect Uber, surely you must acknowledge the possibility that Bloomberg is simply working to undermine them.

Maelochs

This is worse than politics.

Look ... Uber's sensors performed Way less well than most people expected. Legally, the company is not at fault, but its AV system is not nearly as good as pretty much everyone, fro experts to ignoramuses, say it should be.

Idon't see any way any rational person thinks Radar and Lidar shouldn't have picked up a pedestrian in the middle of the road.

In the end, Uber's AV control system is about as good as a human driver----but it should be much better in a situation like this.

If there had been an accident in a dense fog in heavy traffic with drivers switching lanes unpredictably and using the shoulders or the median and reentering the road ... no one would be too surprised---too much data, every possible difficulty, unpredictability, human stupidity .... but on pedestrian crossing 40 feet of pavement .... Uber's system is not ready. Which hopefully will not still be the case when Uber puts its fleet back on the road.

it will be interesting to see what the NTSB has to say.

Otherwise ... enough with the conspiracy BS and the stupid bickering, maybe? Let's just roll with the developing situation and see what tech upgrades Uber offers.

InOmaha

This is much better then politics. We're actually discussing the issues in a reasonable tone and manner while attempting to piece together what actually happened.


Politicians will vote on knee jerk feel good legislation they've never even read to fix the problem. Do something legislation with limited facts, debate, or expertise is their specialty.


That's why I mentioned in the other thread that it's possible people and their proxy politicians could remove bicycles from the roads. That would neatly and conveniently solve this problem going forward and absolve AV manufacturer's of all legal or criminal liability for their designs.


Simply write a law that bans bicycles from the streets and this would have never happened. And if it did, we'll know who's fault it was immediately. It won't be the billion dollar companies who donate to political campaigns or the millions of voters who ride the AV buses.

InOmaha

Maybe one of the problems with the testing platform is they made the car too safe and too comfortable. That quickly lulls the backup driver into complacency. Maybe make one of these self driving so the backup driver has to worry a little more.

KraneXL

Actually it says she was outside of the crosswalk. Even if it detects you in time that vehicle still has to obey the laws of physics. I suspect this may be what happened in this case.

noisebeam

This local news report has some tidbits I hadn't read before:
https://www.azcentral.com/story/mone...ure/446407002/

I-Like-To-Bike

What I believe is that the Tempe police chief showed very poor judgement by drawing her rush to judgement and questionable conclusion from the poor quality dash cam video and the statement of the distracted driver and going on the record with it in a peculiar "exclusive" to an out of town newspaper- “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.”

Her "motive" was probably her own stupidity.

I-Like-To-Bike

Are you now a lawyer, judge and jury as well as the final arbiter of what should or should not be posted on this subject?

Another judge and jury might come to a different conclusion about liability for an organization that puts vehicles on the road and ends up killing people with unsafe vehicles that "are not ready" and inattentive drivers.

I-Like-To-Bike

If the Uber test vehicle was not gathering and recording data from its various sensors, either on board or transmitted to Uber, then what was the purpose for it even being on the road?

I assume that there is recorded data from the various sensors from this vehicle (as well as other Uber "test" vehicles) and that the NHTSA will request that Uber make all of it available as required to make a determination about the suitability of Uber AV vehicles to be "tested" on public roads without real time monitoring by attentive and qualified safety personnel.