njkayaker

The observer was a guy.

njkayaker

The video:

1- there's a car breaking in the distance.
2- her shoes are visible at about 1/4-1/3 into the right lane.
3- The bicycle wheels and her lower legs are visible soon after.
3- Her head is identifiable about 1/2 of the right lane.
4- At around 3/4 across the right lane she turns her head towards the car (before that, it appears she was looking forward).
5- She's on the right side of the bicycle holding something white (a bag?) in her right hand.

squirtdad

some outside "experts" suggest AV should have been able to detect the person killed. ???


https://www.yahoo.com/tech/police-re...232906380.html

njkayaker

Sure. There doesn't seem any reason it shouldn't have been able to.

The video does not necessarily show what a human could have seen.

CliffordK

Part of the problem is becoming "trapped"

It is a common type of accident when one is in the middle of an intersection when the light changes.

In this case, the woman may well have started walking across the street 1/4 mile in front of the vehicle. At which point, there is an expectation that the oncoming vehicle will slow down or stop. Looking away from the vehicle, there is a greater expectation that the vehicle will yield, at least in theory, except when they don't.

I-Like-To-Bike

The dash cam doesn't show too much as it is obviously a POS device, or that is what the released video indicates, but it does show the direction the victim was walking to dispense with the notion that she suddenly emerged from a hidden location on the right side of the road making the collision unavoidable by a man or machine controlled vehicle.

squirtdad

The video clearly show the 'driver" looking down twice once for about 3 seconds and a second time, ended by the crash for about 4 seconds.

whether the driver was looking at a cell phone or something else is not clear, but what is clear is she was not paying attention for about 7 of 9 seconds before the collision

njkayaker

The video might not show everything and it might be giving a misleading impression of how visible she was but the stuff I listed appear to be fairly clear.

njkayaker

Oops. The "driver" is a guy (I associated "she" with the victim).

https://finance.yahoo.com/news/polic...232906380.html

I-Like-To-Bike

True.

squirtdad

confusing as the driver has also been identified as Raphaela...... and the video image could support Raphael or Raphaela

njkayaker

Yes, that's possible. It doesn't preclude her not being careful enough.

It's not a great place to cross.

From around 1/4 to 3/4 of the right lane, it appears she was still looking forward (there's nothing that indicates what direction she was looking at before then).

The video:

1- there's a car breaking in the distance.
2- her shoes are visible at about 1/4-1/3 into the right lane.
3- The bicycle wheels and her lower legs are visible soon after.
3- Her head is identifiable about 1/2 of the right lane.
4- At around 3/4 across the right lane she turns her head towards the car (before that, it appears she was looking forward).
5- She's on the right side of the bicycle holding something white (a bag?) in her right hand.

This doesn't seem to be anything more than a guess.

mr_bill

The safety driver is a woman, Rafaela Vasquez, and prefers the pronouns she/her/hers.

-mr. bill

KD5NRH

Well, it's hard to find light black shirts.

njkayaker

Felon who was at wheel of killer self-driving Uber car | Daily Mail Online

Leisesturm

Of course. THIS, people! Do you all really think the test car was relying on a $19.99 dash cam to do the good work of navigating tens of thousands of dollars and thousands of pounds of test vehicle and human test pilot through dark city streets! Do you? Of course it had other sensors. And likely they detected the jaywalker. And? What? Pointlessly apply the brakes and possibly make things worse? The jaywalker was doomed. The computer knew it. The computer was not programmed for pointless responses! This will be hard for human forensics experts to fathom WHEN this happens again.

I have a cousin who lived for decades in a semi-vegetative state after a collision at night at age 27. He cost us at least $20M in medical expenses and I am not saying it wasn't worth it. I know it wasn't worth it. Had he been able to, he would have said it wasn't worth it either. I knew him well. Top Gun type. Man of action. Smart, funny, active. No way he wanted to exist for 22 years on a hospital gurney with tubes everywhere. I doubt that deer that got bumped at 15mph lived all that well afterward. Or for long.

A human driver wouldn't even be charged in an accident like this. In fact it is a certainty that on the very night this woman was killed at least two other Americans met their demise in exactly the same way except because the drivers were human the non-story failed to make even the local news.

Leisesturm

And ... IDK ... you don't think it likely that a vehicle expected to operate in darkness would have such a mode for its camera(s)? Maybe even something approaching military spec 'night vision'? I do.

njkayaker

No, we don't.

Leisesturm

The tech is there. WE are not there. LAWYERS are why the 'test pilots' are there and GREED is why they aren't paid better and/or vetted better so that they may actually provide some useful purpose. But it really boils down to the impossible Catch-22 that if a human test pilot could intervene then there was no emergency. In an actual emergency even a human is helpless to act fast enough as was the case here. So... no human test pilots necessary. By putting the cars on public roads at all the developers were confident enough in their ability to function optimally. Anyone who thinks this accident could be prevented believes in miracles and is beyond my reach or even my interest in convincing otherwise. There was a time when supersonic air travel was approaching routine. A single accident ended that and decades later there is still no interest in resuming supersonic air travel. There are many who even claim it is impossible. When informed that not only is it possible, it existed. They scoff. Much like many of you scoff at the idea of a computer driving an automobile at all, let alone better than a human. Yes, it is entirely possible that a single fatal accident could bring AV development to a complete halt OR at least result in several jurisdictions in the country refusing to allow their introduction into the traffic stream. Indefinitely. Humans are just that capricious.

Leisesturm

speak for yourself.

jefnvk

She was already into the lane when her feet became visible. The camera shows enough to calculate a close enough estimate of distance, the known speed can tell you how long she had, and she didn't make it across an entire lane. Even watching the video, you can clearly see from the Youtube timer that an estimate of a second is right in the ballpark, although unfortunately the timestamp is obfuscated.

And it may have been. Or a driver may have had crappy eyesight. Or the foreground light she was walking behind may have been brighter and caused too much of a contrast for a drivers eye to adjust. Or a driver may have been looking to the right side of the road where the sidewalk is and not caught her in time. Or any other number of scenarios.

What-ifs can work in both directions. As I mentioned, I have been in this scenario before, where folks in dark clothes were playing in the middle of the street not in view of a streetlightat about 1am. No, people in black clothing in the middle of a dark road where you don't expect them to be, even if driving attentively, aren't always obvious.

Valid point, and it may be a consideration for the future, but again to draw this to human drivers, no one does that. According to this from NHTSA (https://www.nhtsa.gov/DOT/NHTSA/Traf...les/deadly.pdf) we'd need to set all unlit road nitetime speed limits to around 35MPH.

Headlights are notoriously bad on many new cars, it is actually one safety feature I wish would come up to modern day par. People don't test drive cars at night, it is an easy way to cut corners on price. It wasn't even until 2016 that IIHS rated vehicles based on them: Most midsize SUV headlights are marginal or poor. XC90 is at the bottom of the "marginal" rating

njkayaker

I don't think anybody else made the claim.

???

noisebeam

Just out of curiosity I looked at the Strava heatmap of the area. Visible is a very faint track where some cross very near to where she did. The blue box is the intersection of Curry and Mill. The two bright yellow tracks just north and south of the lake (salt river) are MUPs
Capture heatmap.JPG

unterhausen

There is a path there, you can see it just behind the car in the above aerial pic. Not sure why they decided to make a path terminate there with no crossing, it's weird.

There are no cameras with the dynamic range of human vision. Why wouldn't they be using an IR camera, that would have shown the victim just fine. I feel like this might have been a close call with a human driver in most cases. The human drivers looking at their phones would have also killed the victim, but I feel like this is not the standard which a AV should be held. I also am pretty sure there is a programming error exposed by this incident.

jefnvk

Uh, no. There was never a time when supersonic air travel was remotely close to routine. The immense cost of operating supersonic aircraft, the lack of profitability, the fact they were three man flight crews, that they were nearing the end of their service life and there was no immediate replacement or demand, and the operational fact they were limited to transoceanic routes already had supersonic planes on the downward spiral, the Air France crash was merely the coup de gras that did them in.

A more akin situation would be when the first Airbus fly-by-wire demonstration system crashed at an airshow in France in the 80s, we did not stop fly by wire development.