Driverless Cars

[3930dave]

I went looking just for the time of day of the accident, and found the following:

"One of those vehicles was driving north on Mill Avenue at about 10 p.m. on Sunday when it hit a woman pushing a bicycle laden with shopping bags across the street. She later died at a local hospital."

Federal Agencies Investigating Fatal Self-Driving Uber Crash in Tempe | Phoenix New Times

That aspect (bolded) has not been widely reported, and I now understand why there was a damaged bicycle in one picture.

Also:

"Herzberg is said to have abruptly walked from a center median into a lane with traffic. Police believe she may have been homeless."

Uber May Not Be to Blame for Self-Driving Car Death in Arizona | Fortune

So, if that reporting ^ is reasonably close..... let's recap......

Dark, a human is pushing a loaded bicycle across a median onto a road. At a guess, that is likely not a top 10 or even top 100 scenario for the Uber software.

Much larger than human object on a median - autonomous vehicle's vaunted much faster than human reflexes don't do much good, if that lady was classified as a median planter.

Even in urban areas, I'll use high-beams more often than many drivers - may mean little/nothing to Uber sensors, but should help a human driver. Were the high beams On ? I'm not implying that they had to be On, it's just one factor in reaction time at night..... Cell phone records - was the monitor driver on their phone at time, actively, or shopping on Amazon ?........

Plenty of factors play into most accidents, and there is often conflicting reporting....

"Glover downplayed Moir's statements, saying some were taken "out of context" by the Chronicle. The chief disagrees with the Chronicle's headline, "Tempe police chief says early probe shows no fault by Uber," Glover said.

It's too early to say which party was more responsible, he said, and Tempe wants to "wait to reserve judgment.""

^ that quote, is from later in the Phoenix New Times article above..... A Bay Area paper coming out as pro-Uber ? Who da thunk it....

Had this been a senator's daughter attending ASU, who was killed out jogging late at night, would the Blame the Victim vector have shown up this early ?

A guy I went to college with had an elderly gent step out right in front of his car. Knocked him out of his shoes, and killed him on the spot. Some accidents are not possible to avoid; remains to be seen with this one.....

Rgds, D.

 

[Oaktree]

Had this been a senator's daughter attending ASU, who was killed out jogging late at night, would the Blame the Victim vector have shown up this early ?

A guy I went to college with had an elderly gent step out right in front of his car. Knocked him out of his shoes, and killed him on the spot. Some accidents are not possible to avoid; remains to be seen with this one.....

True, but this is a rather big black eye to a technology that seems to be being foisted upon us in the name of "safety". As someone mentioned a few pages back, maybe this can work on special designated routes, but in mixed use with pedestrians, pets, parked cars they've got an uphill battle.

I'm also curious how this technology will handle the challenges of rural driving where blindly following a GPS can get you lost or sent down an abandoned road that's still "on the map" but is now suitable only for off-roaders. What about dirt or snow-covered roads? Does it know about potholes, frost heaves, etc. or will it hit them at full speed?

I enjoy driving. I get the impression that the ultimate goal here is to ban driving as we know it today and force us to use this "for our own good". Hopefully I'll be gone before these cars become the only legal method of transportation.

 

[Fastball]

I have a lot of faith in technology - less so in human beings. While computers will do exactly what they were programmed to do...no computer ever decided to slam the brakes on in front of a semi because he felt tailgated. Yes, there are problems there still...I can just imagine the conversations 100 years ago about the potential perils and “what ifs” about the automobile and the airplane.
Granted, totally autonomous vehicles are probably decades away, but the first steps are already there.
Maybe, instead of spending 10s of billions on far-fetched flights of fantasy like border walls...and 100s of billions of dollars on new, more efficient weapons of war- maybe we could invest in new, modern roadways and infrastructure for the burgeoning technolgies out there on the horizon. .

 

[4570Man]

Driverless cars are going to malfunction and people are going to die. Now let's just sit here and pretend than a human operated car never killed anyone.

 

[3930dave]

Whether I pick up a power saw, or interact with much more complex systems, I want to know what their limitations are.

And..... what risks I'm accepting. Or not.

There are times you shouldn't use a technology. To pick an older/simpler example - Activating cruise control, if roads are icy.

The bar for "improved performance" is actually pretty low.... that's why I'm paying attention..... these autonomous vehicles will be flooding the roads relatively quickly.

A drunk driver can kill one or more people. Technology tends to up the ante..... Imagine the security target a million driverless Uber cars presents ?

Rgds, D.

 

[3930dave]

I'm also curious how this technology will handle the challenges of rural driving where blindly following a GPS can get you lost or sent down an abandoned road that's still "on the map" but is now suitable only for off-roaders. What about dirt or snow-covered roads? Does it know about potholes, frost heaves, etc. or will it hit them at full speed?

The current tech seems to have an issue with Big items. In just this little thread, we have examples of hitting the side of a transport trailer, and the rear of a full-sized fire truck - both at significant speed.

Maybe the smaller stuff is OK ? :confused3:

Rgds, D.

 

[sd455dan]

Driverless cars are going to malfunction and people are going to die. Now let's just sit here and pretend than a human operated car never killed anyone.

True, .
But
what may come out of this sad event may be more overlap and redundancy in different sensors -guessing infra -red, night vision- possibly more advanced lidar may all need to be used in addition to just regular (vision) to reduce another occurrence. It may be true that it was an unavoidable accident, but I wonder if the person with bike that was hit and killed had crossed that same place at night other times and had been easily seen and avoided by a human driver.
I also wonder if the uber moniter will change professions after this. Can't imagine that this will not permanently affect them mentally. The whole (idea) of trusting the car and it's systems to not need to be ever vigilant while sitting behind the wheel is the problem imo. at least for the near future.

 

[crazyal]

The video has been released. While it shows the woman just appearing out of the dark while the person who was suppose to be ready to override the computer was distracted it's a low quality video. The problem is that the video may make it look darker or harder to see the woman than it really was. The real problem is the video is just for reviewing, the actual sensors never picked up on the woman and the car drove right into her. Not only is this a black mark for driverless vehicles it really brings into question if they really are ready for use on public roads. Day and night shouldn't make a bit of difference to the sensors. Even if they didn't didn't pick up on the woman in time to stop (but should have) the vehicle didn't attempt to emergency brake or swerve as a human driver would have.

Twitter

 

[3930dave]

Average crowded city streets...... people step out from in between large parked trucks, construction dumpsters etc..... all the time...... add cell phones (a form of population reduction....)........ and, it's not just the "kids" doing it......

I view this accident scenario as commonplace (in cities at least).

Rgds, D.

 

[3930dave]

The video has been released. While it shows the woman just appearing out of the dark while the person who was suppose to be ready to override the computer was distracted it's a low quality video. The problem is that the video may make it look darker or harder to see the woman than it really was. The real problem is the video is just for reviewing, the actual sensors never picked up on the woman and the car drove right into her. Not only is this a black mark for driverless vehicles it really brings into question if they really are ready for use on public roads. Day and night shouldn't make a bit of difference to the sensors. Even if they didn't didn't pick up on the woman in time to stop (but should have) the vehicle didn't attempt to emergency brake or swerve as a human driver would have.

Twitter

From earlier reporting, I was expecting to see someone stepping off the median immediately into the path of the car. An average driver with high beams on never would have made that mistake.

Rgds, D.