mr_bill

We also need better training for people who ride bikes, such as people on bikes yield to pedestrians. (Even pedestrians who are not in a crosswalk.)

-mr. bill

CliffordK

We don't know what the driver was actually doing before the crash. Apparently there used to be two drivers, one to interact with the computer, and one to watch the road.

Uber recently reduced that to one for both tasks.

So, it is quite possible that the driver was watching the computer display. Perhaps the computer was giving information like identified street sign, lane borders, etc.

robertorolfo

What are you talking about? We are taking a real life event and we are using it to discuss the larger issue at hand. The concept really isn't very difficult.

What we have seen is that an AV from a large company, and which was allowed to "test" on public roads, completely failed to function as designed and as promised. It simply didn't work.

So don't tell me that you know AV's will detect lane taking cyclists, because this one was supposed to detect a J-walker in that situation, and it didn't. Don't tell me that you can trust any AV that is still in the testing phase to be used on public roads.

Testing implies a risk of failure, and the stakes are simply too high.

noisebeam

I am talking about your hypotheticals to which I responded. Why ask about hypothetical if you were intending to talk about real life event?? I responded to this:
It did not completely fail to function as designed. It failed at a type of detection and/or response.

I noted there are AVs today that detect lane taking cyclists very well in a wide, but not complete, range of circumstances. There may be cases they don't very well, but that is why we verify.

Maelochs

Ah, Noisebeam ... just because you ride a bike in Tempe, in traffic, with other cyclists, human-driven cars, and AVs ... doesn't mean you know anything about riding a bike in Tempe, in traffic, with other cyclists, human-driven cars, and AVs.

We are in the presence of the omniscient genius. Your real-world experience ... nah, just not relevant compared to the imagination of an omniscient genus.

noisebeam

Possible, but if you watch the vid you will notice she had a little smile/laugh the first time she looked at device. I doubt routine info about the job at hand would cause that response.

Maelochs

"Wow, I have never seen this warning before ... 'Approaching illegal pedestrian', what could that--Woah!!!"

CliffordK

What do you think made her eyes pop open as big as grapefruits?

McBTC

If you mean to be playing the part of a raving alarmist for whom facts don't matter, good job. If the system was turned off, then we could be looking at the situation, from among myriad possibilities, of a distracted driver who might have well been driving with simple cruise control activated, and a pedestrian with diminished mental capacity who ambled out in front of a moving vehicle, suddenly appearing out of nowhere in the dead of night.

robertorolfo

Wow, are you really struggling this much? These are hypothetical scenarios and questions based on a real world example. Is it really so hard to understand? Are you familiar with the scientific method?

Yes, this AV completely failed to function as designed. There is no question of that. It was designed to detected objects in its path, and then to respond accordingly. It either did NOT detect this person, or did NOT respond accordingly, or both. It failed to work as intended. Simple. Fact. Done.

And yes, you test things to "verify," but those tests don't have to put people at risk, do they?

What does that have to do with the issue at hand? Nothing.

Raving alarmist? Did you see the video? Did you see the part where the AV completely failed to notice a human in the road and plowed right into them without hesitation?

And if you want to call people names, your despicable (and irrelevant) comments about the person who got hit would make you what, then?

Rollfast

Hey kids, I just got to the playground and I didn't buy the fight program?

McBTC, your language got stuck in the mud bog and I don't even know why yet.

If this is really getting to you, maybe SBW is right.

Good GRAVY.

BillyD

Let's lighten up on the name-calling and insults gentlemen.

02Giant 

Absolutely.
Uber paid off the family to cover up it's shortcomings. This has to be the fastest pay-off I've ever heard of.

RFEngineer

Exactly. This seemed awfully fast for a settlement to be reached. But I guess Uber is desperate to keep the family quiet and make this go away in general.

kylea

poor uber

unterhausen

merged the three threads on this subject

jon c. 

Why does that matter?

KD5NRH

Given that no parachute was found, I think we can exclude a HALO drop. That's the only way, by current technology, that she could have appeared where she did without having been in the LiDAR's view for roughly 3.7 forevers worth of computer processing time.

And did we forget these cars are also supposed to be equipped with forward radar? Are bicycles and aluminum cans the newest stealth technology?

KD5NRH

Ever driven on a country road at night? If the low beams can't handle even 38mph, how is it that people routinely survive driving 70+ with them?

KD5NRH

Well, since we conveniently never get to see the two videos synced up, it's safe to assume whatever it was isn't helpful to Uber's cause. Same goes for the tons of other data the system should have been collecting.

McBTC

Jaywalking is a choice and there may be consequences. A homeless person probably has made a lot of bad choices throughout life. One of the consequences is possibly being hit by a car--e.g.,

and,

McBTC

The CDC says that about, "129,000 pedestrians were treated in emergency departments for non-fatal crash-related injuries in 2015" Anyone jaywalking a darkly lit road late at night -- not even wearing reflective clothing -- knows they're engaging in a dangerous and possibly self-destructive act. How many times have we heard this when growing up?

02Giant 

Uber is more desperate to keep their AV records hidden from discovery. I hope this tactic fails.

genec

Yes, yes, we are all aware of all that... but we are also aware of the supposed superior vision and reaction speed of AVs... so why didn't the AV react?

Is it programmed with the above laws in mind, to hit objects, such as people, or bikes, not in compliance with jaywalking laws?

njkayaker

The AV systems are supposed to be able to deal with this easily.

This wasn't an "edgy" difficult situation.

The AV didn't even slow down with the person in the lane.

It appears the AV system handled the situation worse than a careful/attentive driver would have.

This is kind of gross.

Why do you think the "bad choice" of the pedestrian is enough to give the vehicle a pass for an incompetent response (which is a different "bad choice")?

=====================

In accident analysis, the ill-outcome is often the final result of a chain of multiple errors. That is, rather than an appropriate/safe reaction to the initial mistake, people keep making additional mistakes.

That is, often, there are multiple opportunities to correct and avoid the end result but none of the opportunities are taken.

=====================

In this case, the sequence of mistakes appears to be:

1- Poor design of AV system (speculative).
2- Pedestrian walking in an inappropriate place.
3- AV systems not detecting or reacting appropriately.
4- "Safety" driver not being attentive.

End result: a (likely) avoidable death.

Whatever the cause of 2 might have been, it doesn't free the people behind the mistakes in the other steps from responsibility.

That is, those people behind the other steps are required to take due care to avoid the other mistakes even if 2 never occurred!

We only found out about the other mistakes because somebody happened to die.

Given the apparent poor performance of the AV system, it doesn't seem it would be able to handle a pedestrian crossing at an appropriate place.