noisebeam

The dirt path there has been discussed before. The path was not built there, but result of usage ('desire path' is what they are sometimes called). However there really is little advantage to crossing there vs. going to intersection just north or even better under Mill which is what I usually do. One has to make the North/South travel at some point so better to do it before or after the road than to cross illegally - in other-words crossing mid block may feel shorter, but in reality one will walk the same distance.

Rollfast

Supersonic air travel ended because it was expensive, the planes (Concorde SST) were very noisy and needed special conditions to arrive and depart (and they polluted like all get out) so NYC was about the only city it came to in the US.

The Concorde was a creation of the 1960s, with a lot of innovations no other passenger jet had to deal with. I think it had to do with parts more than an accident. The newest idea is to make a partial ORBIT and come down in the fashion a spaceship might.

Supersonic passenger transportation is just not something the average flyer can afford.

Rollfast

Rafaela is pretty much a female name. I didn't see where the confusion came from.

njkayaker

That wasn't the name use initially (by some places).

unterhausen

"Uber's cars have killed people at roughly 25 times the rate of a typical human-driven car in the United States."
https://arstechnica.com/cars/2018/03...s-car-program/

02Giant 

But, she shouldn't have been in the road...

Rollfast

Kinda hard for them to quantify that or argue if the others were actually driven by humans...I didn't read but somebody sounded mixed up writing that.

I still want another Duster before I'm dust.

Leisesturm

My point isn't to take is into why or why not we can't have supersonic travel so much as to illustrate how risk averse to progress the general public is. Fast transport is definitely useful and could be affordable. AV cars can be useful but a majority of the general public are looking for any reason to say they should not be explored any further. Look around at all the stalled R&D projects over public squeamishness or fear. There is much more fear in this thread from detractors than there is encouragement from forward lookers. I won't even visit that other AV thread anymore, its just too depressing.

The car was outdriving its lights... it was going 38 in a 45? Would a human be criticized that way? And please stop trying to compare human eyesight to the suite of sensor inputs of the AV. Of which sight IS one, and for a certainty it is better in all light conditions than the best human airline pilot. Count on it.

The Space Shuttle Challenger blew up and there was an immediate hiatus (10? 15? years) to further missions. Then we tried again and Columbia blew up from the same cause! And now America has no way to reach near Earth space except to pay the Russians $75M (per head) for a ride in their Soyuz equivalent. I don't think there will be a 15 year hiatus before AV's can start testing again but I know there has been heartfelt negative reaction to the news of a human being killed by an AV. Pure overreaction. Why were Americans allowed to even think that an AV could NEVER crash? NEVER kill someone? They are going to do plenty of both! About the only reason why this accident will probably not be the END of the AV in our lifetime is that first blood was that of a (presumably) low net worth individual.

noisebeam

This is not true unless the car did not have standard issue headlights or they were defective.
If it was true than no car with headlight meeting standards could safely travel over ~35mph on an open unlit road.

KD5NRH

Mil-spec night vision would be washed out by the headlights. Anything with a decent low light mode would be showing a lot more, even if the car was only using parking lights.

The front view shows roughly the illumination area of a set of pre-halogen sealed beams. Unless they went digging through antique auto parts catalogs to fit out their AV, that camera is nowhere near as sensitive as normal vision.

And considering the whole point here is supposed to be showing that AVs are safer than humans, whether the human reacted in time or not is irrelevant to the fact that the AV didn't. And given that AVs have no inherent visible spectrum limitations, there's no excuse for it not seeing the woman in the middle of an open, relatively flat street long before the human could have.

KD5NRH

An alert driver, yes. Even if the babysitter hadn't been distracted, though, the time to get hands on the wheel and a foot on the brake might have been too long. Still 100% an AV fail.

Road Fan

I would say the vehicle should also have been designed to detect and identify vulnerable road users in a wide variety of external conditions. Failure to detect leads to failure to act, so it is an unacceptable failure.

squirtdad

I have watched the video a few times......I am becoming of the opinion (not right of the opinion) that an alert driver or AV could have swerved in that situation and avoided the collision.

KD5NRH

How long do you think you can sit holding your foot poised over the brake pedal? Also, since most of the AVs have sensors to detect any input, no matter how small from the steering wheel, they're likely told not to touch it except when it's necessary to take control.

noisebeam

For perspective AZ has the highest pedestrian fatality rate in the US.
https://www.azcentral.com/story/news...ays/383640002/

In Tuscon 70% of pedestrian fatalities occurred during mid-block no crosswalk crossings. 70% were also at night. I suspect this breakdown is similar across other AZ metro areas.
Road Runner: Tucson looks to make pedestrians safer | Local news | tucson.com

In general in AZ metro areas speeds are too high and there are too many lanes to cross quickly.

Statistically ~2.5 pedestrians have been killed by autos in AZ since Elaine was on Sunday night.

CliffordK

According to that list, some volvos are high, some are low depending on the year and model.

I was thinking about lights some. For a computer controlled system, right dipping is only necessary on a 2-way road. On a one-way street, symmetrical dipping could be used. So, perhaps an AV could have custom headlights that would change based on where it was.

But the other point is that an AV should be able to use headlights in multiple bands, from UV to IR. Bright white light may, in fact, be problematic for an AV. However, one should be able to strobe headlights well above flicker fusion. Then time the camera captures over different bands based on timing.

Of course, get a bunch of these cars on the road, and it could be problematic unless one would use a universal timing such as tying the vehicles and perhaps overhead lighting to GPS clocks.

Leisesturm

You missed my sarcasm. An earlier poster claimed that the car was outdriving its lights. I agree with you.

Leisesturm

Seriously, y'all's make a very good case for just scrapping the whole AV thing and putting up with the status quo. The bellyaching over inevitable negative encounters between AV's and humans is getting annoying. This should have been, that should have been, the other should have been... ... the coulda, woulda, shoulda, is so thick up in here it's getting hard to see the reality. What I KNOW is I would not have wanted to be the pedestrian that this car 'saves' by hitting me at 20mph instead of 38mph! There really is worse than being killed instantly. Didn't any of you read my earlier post? 22 years as a quadriplegic with the mental function of a newborn baby is NOT what I'd call being 'saved'.

Leisesturm

So could I. The test pilot needs to be jailed for a very long time. Maybe for life. Know what? IF we had the stones to jail drivers that kill pedestrians for 25 to life, with no exceptions ... we wouldn't need self-driving cars! In 5 years ... less ... the vehicle death rate would drop to EU levels, and the R&D hassle of getting the Level 4 AI to really do this stuff right would not be worth the headache.

noisebeam

This happened not to long before and not to far from this incident and no one is talking about it
Police search for suspect who threw metal object into a bicyclist's spokes, killing him - Story | KSAZ

CliffordK

I would suggest at this time that all Uber self-driving vehicles be temporarily pulled from the road.

Then I would suggest what can be called the "Tempe Test" to get them back on the road.

For the Tempe Test, Uber's CEO and lead project engineer must return to this same Arizona crossing and walk across the road in similar lighting conditions, timed for the Uber car to approach them at 38 miles per hour to meet them in the same lane.

And, of course, also prohibit VW style test modifications to design the car to only pass on certain dimly lit Arizona streets (perhaps also choose another random place for a second similar test).

Perhaps a similar test would be light timing, and hitting a red/green light change while a bicycle has been caught still in the middle of the intersection. Does Uber predict light changes?

Not meaning harm to those individuals. But, they should be able to assure the public that they have enough confidence in their cars to walk in front of them under mixed lighting conditions. And they are no better or more important than other street users that the cars should be protecting.

As cyclists and cycle commuters on this board, we put our lives on the line every day. And, the Uber isn't necessarily a car I would like to share the road with.

I need to track down the first incident. But, I wonder if the incidents relate to a faulty right-of-way algorithm. So, if the uber car thinks it has the right-of-way, it just goes straight full steam ahead, and others should get out of its way. Ok, perhaps not quite that extreme, but it may well give itself too much priority.

Another issue, of course, is object permanence vs movable objects. Did the car detect the pedestrian early in the left lane, and not realize she would move over into the right lane? Seeing a pedestrian already committed to being in the road should be a huge red flag (although some pedestrians will stop in center turn lanes and median strips, but they still should garner extreme attention). Baring a center turn lane, a pedestrian in the road should mean they are committed to crossing. Even if the pedestrian pauses for a car to pass, the pass should be done slowly.

CliffordK

What about Uber? They obviously had a look ahead camera that could check up on their drivers.

It wouldn't take much to write an eyeball & gaze algorithm that would let them sort through thousands of hours of tapes to see which drivers were taking their jobs seriously, and which weren't.

Where were the driver's eyes going? Digital readouts on the car? Some other device, or nodding off?

Anyway, those that weren't doing their jobs should be terminated and get drivers motivated to actually take it seriously.

02Giant 

I get that, coming from someone that wouln't look at the big picture.

CliffordK

Or simply permanently revoke the people's driver's licenses. Then those that get caught driving get sent to the slammer.

For some, no license could be a fate worse than death.

02Giant 

I had thought the same.

A few here fail to acknowledge that. "She shouldn't have been in the road"