njkayaker

Based on the video I posted 36 minutes before your post, it appears you are incorrect.

rgconner

Most "devil's choice" questions rely on having perfect knowledge about the situation.

As a result, such dilemmas don't work well in the real world as we never have perfect knowledge.

A logical error known as the Horns of the Bull, or a false dilemma.

njkayaker

Dunno. "Many mistakes were made."

CB HI

Thanks for the video guys. BFs or my computer were acting up and did not update to include the post with the video.

It will be interesting to see if the police still want to claim the Uber car and back-up driver had NO fault.

Pretty clear why Uber shut down testing so quickly.

Maelochs

One second from the time I see her shoes (and only because I know what to look for) and the impact. Don't know why radar and lidar didn't work. Don't know why the ladies eye and ears didn't work.

I know that guy won't be working long---looks like he was texting, picking music, or just nodding off at the wheel.

One thing I do think, though ... if it had been me at the wheel, that lady would be in the same shape. Seriously---- headlights to impact was one second, and as best I know, reaction time is at best half a second, and at 55 feet per second or so ....

But .... that is one of the kinds of things AVs are supposed to prevent with all their sensors.

Oh, and for all the people saying this footage wouldn't be out for years and would be heavily redacted .... ********** <crickets>

njkayaker

It appears that a "backup driver" wasn't a requirement.

CB HI

Remember, video resolution at night is extremely poor. The human eye can see much better than the video implies was available to the driver, of course that means the driver had to be paying attention.

Under similar lighting and speed at night, I have spotted a shadow off to the side and slowed in time to prevent hitting a homeless drunk.

prathmann

Yes, I trust the other sources more on the speed limit. And after seeing the video referenced above I also disagree with the police assessment that the crash was unavoidable. In the video it is hard to spot the pedestrian & bike until too late. But my experience with recorded video vs. live driving is that I can see far more outside the hotspot of the headlights with my eyes than I can on the video (human eyes have a very wide dynamic range). Furthermore I usually use my high-beams under those conditions - wide center median and no immediate oncoming vehicles.

I doubt that a human driver of a car under these conditions would face any charges since the pedestrian is clearly crossing improperly and without even glancing for oncoming traffic. But I would have expected more from an AV with additional sensor capability such as LIDAR.

njkayaker

https://www.theguardian.com/technolo...d-by-kangaroos

I-Like-To-Bike

That's right, blame the victim because the AV took no evasive action despite what the video appears to confirm from the photo graphic from the NYT. The victim appeared to walk at least 30 feet of the highway unobstructed from the view of the Uber vehicle which struck her and the bicycle anyway without taking any evasive action.

The Uber car software/hardware either didn't "see" anything or the hardware did not respond in any way to what the software did "see."

I have to wonder what video the Tempe police chief looked at before making his mystifying and exclusive statements to the SF Chronicle

njkayaker

I didn't blame the victim. That's just you putting words in my mouth.

The highway is unobstructed from the view of the pedestrian too.

Do you think she saw the car?

gpsblake

Those automated cars have a higher standard to meet than a human driver. A person crossing the road the way that woman did at night and considering the highway, I can see how easily it would be to hit her. However, what does bother me is that the car made no effort to slow down at all before striking her and should be able to see where we can't. Did the car have infra-red or does it just rely on what we use for sight? Maybe putting infrared in these cars at night might be a good idea.

350htrr

Why didn't human drivers "see" people they drove over... Thousands and thousands of them, for the last 100 years or so... In the end, things will get better, as AV WILL improve, unlike human drivers, every generation starts from the bottom and kills people the same way, over and over and over...

I-Like-To-Bike

If your night vision or the Uber AV sensors are as limited in range/reception as the camera used by the Uber neither you nor the Uber vehicle should ever drive at night.

I-Like-To-Bike

Don't know and it is irrelevant and why do you keep harping on whether the victim saw the Uber, unless you are trying to shift the topic to blaming the victim?

The issue is that the Uber AV system apparently failed miserably and the woman is definitely dead at least partly as a result of the inability of the AV to take any evasive action prior to colliding with an unwilling unpaid test subject of the Uber AV testing program being conducted (apparently unsupervised and mostly unregulated by AZ or local officials) on public roads.

Daniel4

I viewed the video several times especially the point where the woman first became visible.

I wonder what would have been different if I were driving. It's difficult from the video. A lot of people say the car was going 40 in a 35 zone. Generally, nothing out of the ordinary. But it looks as if 40mph is driving beyond the lights. A good human driver would observe the 2 second rule not only for following the vehicle in front but to not overdrive the lights. (And in the process, get passed by an impatient driver - so the woman would have been struck by somebody else.)

Having said that, the car's technology should have picked her out. That's what we expect from this technology. However, if AVs are to rely on human back-up, then it must drive within the 2 second rule to allow the backup to react and override.

In this case, the backup never had a chance even if she was paying attention. Maybe additional systems should have an operator attention sensor.

I-Like-To-Bike

Maybe because they (or their associated software) don't always work as well as promised in the promoters' press releases.

I-Like-To-Bike

Hasn't Microsoft offered versions of Excel and Word for the Mac since the 1980's?

Rollfast

Not anymore.

I-Like-To-Bike

What is this this if not Word and Excel for Macs? https://products.office.com/en-us/ma...office-for-mac

SHBR

Well, I might add that if you have a fatality on your hands because of a trade secret, it might not be a great long term strategy.

This day was coming, in some ways this could be positive, if everyone looks at all of the details closely and tries to prevent this kind of incident in the future.

There is a video. Looks quite tragic.

Too many blind followers in society, not enough people asking questions, the bridge in Florida is another example of this mindset.

SHBR

Elaine Herzberg did not deserve to die.

Shame on anyone who thinks so.

I have to agree, this was preventable, it calls into question everything about AV technology.

I-Like-To-Bike

Are you still sticking with that story?

tyrion

Plenty of people are asking questions. I don't know why you'd assume otherwise.

I-Like-To-Bike

Cut and dry IF the Uber vehicle is being controlled by sensors that only "see" what is in front it as poorly as the Uber video camera records what it "sees".