Experts: Uber self-driving system should have spotted woman

AP ^ | March 22, 2018 | TOM KRISHER and JACQUES BILLEAUD

[Former Proud Canadian]

No exerpt.

Video of accident at link.

[DiogenesLamp]

When driving on a well-lighted roadway? Mebbeso...

I have read that it was a well lighted road, but the video I saw did not appear to demonstrate that it was a well lighted road.

I drive down a well lighted road quite often, and I can see someone standing in the road from half a mile away.

[logi_cal869]

Ping

[dragnet2]

That's only 1.166 seconds. Even with vision equal to that of the camera, I seriously doubt that a human would have perceived a threat until much of the woman's legs were illuminated

Ya mean when the victim stepped out of the black, ink like darkness? Btw, any defensive alert driver would have been standing on the brakes at point of impact.

Did the Uber vehicle brake before point of impact? And why would they not tell you this? What happened post impact? What did the vehicle do, what did the occupant do? After impact, did the Uber driverless car run over the victim?

They have all the data. Have you seen it?

[redcatcherb412]

Anyone know if these self driving cars have infrared emitters on the cars? Most digital cameras have infrared capability so the pedestrian in dark areas out of headlight view could be in camera view.

[dragnet2]

Not sure, but I think they're using ToysRUs spy patrol cameras. This is being put together by companies that stand to make hundreds of billions and it's being tested by a billion dollar company.

Lets hope the technology they're using to detect distances, objects and movements are better than what they're using for their exterior monitoring camera. And they only use one forward looking camera? With so much riding on this, that seems a bit odd.

I'd like to know if their super technology braked prior to impacting the victim and what occurred post impact?

[TXnMA]

The bicycle pusher was crossing just outside the limit of, apparently, the last streetlight. If I'd been driving, I might have thought to switch to high beams as I approached the dark area.

Maybe not; in a lighted area, at times, it's difficult to tell if your headlights are on --much less, whether they are high or low -- unless you go "eyeballs in" and check the instrument panel... (Like the "human driver" was -- until the last instant.)

Astronomers would say the cyclist was "just beyond the terminator"... '-)

[DiogenesLamp]

I think a system that relies on video detection of obstacles in the road is defective. I do not think the current state of video detection (unless you are using a time of flight camera) is capable of doing the absolute bare minimum necessary to prevent a collision with objects in the road.

I think the technology they are using is insufficiently advanced to meet the minimum requirements that would be necessary to do this sort of thing safely.

If that thing can't see 300 feet in front of it, it's not good enough to be on the road.

[vooch]

Someone filmed the same stretch of stroad last night with a better camera . Very illuminated

https://youtu.be/CRW0q8i3u6E

negligent homicide

[TXnMA]

Nope, haven't seen it -- and this is my last post on this subject; I have something much more important for FR I'm writing HTML code for right now.

This thread is getting circular, and I haven't time for that.

However, you'd do well to read the article I linked to in #104...

Fregards...

[dragnet2]

The bicycle pusher was crossing just outside the limit of, apparently, the last streetlight.

Look at the video again. The vehicle was between street lights as was the victim, with what appeared to be the last street light on the right, in front of vehicle coming up on the right.

[dragnet2]

Those are some odd comments. My questions and comments were based strictly on the video evidence at hand. Not some professor with a obscure website. He'd no doubt agree with me anyway.☺ Btw, you seem angry. Cheer up.☺

[Former Proud Canadian]

"Did the Uber vehicle brake before point of impact?"

No.

"And why would they not tell you this?"

They did.

[Former Proud Canadian]

"any defensive alert driver would have been standing on the brakes at point of impact."

BINGO!

[Former Proud Canadian]

Re post 123: Yes, yes, and yes.

[EVO X]

Interesting article. It doesn’t say what frequency it works at. Here is an other interesting article about people radar. It would obviously have to operate a frequency such that the signal doesn’t get scattered or absorbed by the structure.

[EVO X]

Birds and raindrops give a radar return.

[GingisK]

The military one was a Doppler Effect device. The operator wore headphones and listened for swishy noises. People sounded like someone walking through Styrofoam peanuts with other odd noises in accompaniment.

It operated at a number of frequencies just to optimize reflections from watery objects such as people.

[GingisK]

A Doppler radar could hear the spokes and her legs moving.

[EVO X]

Doesn’t sound very accurate, but close enough for government work.

[Fresh Wind]

negligent homicide

Thank you for posting that.

Those saying the Über video was doctored are probably right. They are doing an emergency CYA, trying to direct the blame to the road conditions and away the failure of their technology to detect a common problem that drivers face on the road, and the failure of their employee to do the job "he" was being paid to do.

With several seconds of a clear view of the road ahead, there's no way a reasonably alert human wouldn't have seen the woman walking her bike from the left across three traffic lanes.

A diagram of the accident scene appears here, and also here.