That Uber AI Driving test crash.

[bustr]

Another report confirms the driver was streaming "The Voice" when the crash happened. The AI and software is not ready for prime time and Uber thought they had a responsible person in the car. Kind of a Catch-22 oxymoron in that the reason for developing this kind of AI is becasue of people like the supposedly responsible back driver.


http://www.latimes.com/business/autos/la-fi-hy-uber-self-driving-death-20180622-story.html

https://www.theguardian.com/technology/2018/jun/22/driver-was-streaming-the-voice-when-uber-self-driving-car-crashed-say-police

[MiloMorai]

Thought this might be a good thread to post this in.

Driving aids such as lane assist - if one has the windshield replaced the camera mounted to the windshield has to be re-calibrated. A driver in Newfie almost had a head on because the car wanted to go left all the time.

[Karnak]

I am just going to repost what I said on Ars Technica in response to this, well, slightly edited:

Uber was averaging one driver intervention every 19 hours. While the safety driver ought to have been trying to pay attention, it is impossible for humans to maintain attention for that kind of duration when we aren't actively involved or don't find the subject significantly engaging. Watching a road go by is not actively engaging.

The whole Level 3 concept of "The car will drive, but the human in the front left seat will be paying attention and ready to take control if need be." is a broken concept that replaces actual humans and human behavior with theoretical perfect humans. People just don't work that way.

Cars need to go from Level 2's "The car will monitor the distance between it and the car in front of it while the driver steers." to Level 4's "The car drives itself and no human attention is required." Level 3 does not function when human nature is accounted for. In practice Level 3 will be treated as if it is Level 4 and accidents and fatalities will be the result.

[Wiley]

The whole Level 3 concept of "The car will drive, but the human in the front left seat will be paying attention and ready to take control if need be." is a broken concept that replaces actual humans and human behavior with theoretical perfect humans. People just don't work that way.

And the fact the people implementing it don't seem to see that is kind of staggering.

Wiley.

[pembquist]

Some of them don't give a ............

[DmonSlyr]

I am just going to repost what I said on Ars Technica in response to this, well, slightly edited:

Uber was averaging one driver intervention every 19 hours. While the safety driver ought to have been trying to pay attention, it is impossible for humans to maintain attention for that kind of duration when we aren't actively involved or don't find the subject significantly engaging. Watching a road go by is not actively engaging.

The whole Level 3 concept of "The car will drive, but the human in the front left seat will be paying attention and ready to take control if need be." is a broken concept that replaces actual humans and human behavior with theoretical perfect humans. People just don't work that way.

Cars need to go from Level 2's "The car will monitor the distance between it and the car in front of it while the driver steers." to Level 4's "The car drives itself and no human attention is required." Level 3 does not function when human nature is accounted for. In practice Level 3 will be treated as if it is Level 4 and accidents and fatalities will be the result.

Yup. This is so true.

I find this to be extremely scary in the sense that peoole will be trusting computers for action and not themselves. It's kind of like how computers memorize your passwords, and then you forget it when it wants you to create a new one. The same concept will happen with cars.Then "for the sake of safety" they will force everyone to get a driverless cars, which will stricktly impead travel and freedom. It's going to put thousands of truckers out of business. I just don't like it all.

[mikeWe9a]

I am just going to repost what I said on Ars Technica in response to this, well, slightly edited:

Uber was averaging one driver intervention every 19 hours. While the safety driver ought to have been trying to pay attention, it is impossible for humans to maintain attention for that kind of duration when we aren't actively involved or don't find the subject significantly engaging. Watching a road go by is not actively engaging.

The whole Level 3 concept of "The car will drive, but the human in the front left seat will be paying attention and ready to take control if need be." is a broken concept that replaces actual humans and human behavior with theoretical perfect humans. People just don't work that way.

Cars need to go from Level 2's "The car will monitor the distance between it and the car in front of it while the driver steers." to Level 4's "The car drives itself and no human attention is required." Level 3 does not function when human nature is accounted for. In practice Level 3 will be treated as if it is Level 4 and accidents and fatalities will be the result.

The catch 22 in this is that, without sufficient experience in developing and operating the level 3 automation, manufacturers will never be able to get to level 4.  The systems have to demonstrate a long period of reliability in operation, and that period of reliability will greatly exceed the ability of a human operator to monitor it adequately in case it DOES fail (and it will, at some point).  Driving may be simpler than flying from a procedural standpoint, but is generally a more complex and more quickly changing environment.

Mike

[Karnak]

The catch 22 in this is that, without sufficient experience in developing and operating the level 3 automation, manufacturers will never be able to get to level 4.  The systems have to demonstrate a long period of reliability in operation, and that period of reliability will greatly exceed the ability of a human operator to monitor it adequately in case it DOES fail (and it will, at some point).  Driving may be simpler than flying from a procedural standpoint, but is generally a more complex and more quickly changing environment.

Mike

I have no problem with them testing and developing them using public roads, given they are providing stats to the governments so that informed policies can be set and/or developed.

What I don't want to see them do is be sold to the public. 

[John Galt]

Self driving Ubers have a high potential to become self aware and attack their human controllers.

[Oldman731]

Self driving Ubers have a high potential to become self aware and attack their human controllers.

Except in Philadelphia.  They know how we treat robots here.

- oldman

[Hajo]

Computer controlled automobiles,and trucks on public highways is the dumbest thing I've ever heard of.  I worked with programmable logic controlers for well over 30 years.

Most of the time PLCs worked fairly well.  However we (the program controllers and electricians quite frequently had to "fix" and reprogram controllers).  It worked well in manufacturing because when it failed it didn't kill or main anyone. Economically it was feasible.  Sometimes it would take a good deal of time to find, diagnose, and repair.  PLCs are complicated.  However imho to let these logic controllers let loose on the general public where lives and injuries can be the result is utterly stupid.  I know.  I've been there and they failed in hvy manufacturing.  Now you wish to put humans at risk knowing full well "it ain't gonna work right all the time?"  Is there an acceptable rate for injuries or deaths while using PLCs to deliver human beings to there destinations?  I would never, ever get into any vehicle that is dependent upon an inanimate object for my or my families safety.   And any one who is supposed to watch is likely bored to death after about 10 minutes and probably brain dead from the inactivity.  Stupid.  Unreliable.  Can't be trusted fully.  I'll drive myself thank you.  And ya I know human drivers aren't the best but you can't tell me a percentage of injuries or fatalities in this case is acceptable.  It will not be 100% reliable.  It never will be.  Count me or anyone I know out.

[Wiley]

The catch 22 in this is that, without sufficient experience in developing and operating the level 3 automation, manufacturers will never be able to get to level 4.  The systems have to demonstrate a long period of reliability in operation, and that period of reliability will greatly exceed the ability of a human operator to monitor it adequately in case it DOES fail (and it will, at some point).  Driving may be simpler than flying from a procedural standpoint, but is generally a more complex and more quickly changing environment.

Mike

Agreed.  But step 1 should be there should never be only 1 person monitoring the vehicle.  ESPECIALLY if they're supposed to be also monitoring the telemetry.  The fact that Uber apparently didn't understand that is an enormous indicator of their competence.

Wiley.

[Karnak]

Is there an acceptable rate for injuries or deaths while using PLCs to deliver human beings to there destinations?

Yes.  Less than human drivers would have injured or killed.

Waymo is likely not very far from accomplishing that.  Where Uber has a safety driver intervention every 19 hours in easier to handle Arizona, Waymo has a safety driver intervention once every 5000 hours in California and GM has a safety driver intervention once every 1000 hours in difficult San Francisco.