Self Driving Uber Fatality - Thread IV 7

[IRstuff]

This thread is continued from thread815-438410

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[IRstuff]

They are less safe, in the sense that they have a feature called "Autopilot" that people think is an actual automatic driving mode, when it's not even close.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[RVAmeche]

That is true

 

[JStephen]

The reason you hear about the Teslas is not because they're Teslas. It's because "robot kills master" is a much more compelling headline than "man does something stupid and dies". The autopilot causing an accident makes it distinctively different from every other auto crash, regardless of actual safety statistics.

 

[IRstuff]

The autopilot causing an accident makes it distinctively different from every other auto crash, regardless of actual safety statistics.

Because, in many, if not most, cases, it was being MISUSED; the driver is either inattentive a little, or enough to not even be looking at anything car-related.


TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[ImminentCollapse]

"it was being MISUSED; the driver is either inattentive a little, or enough to not even be looking at anything car-related. "

That brings us full-circle. In the 6000+ ped deaths cited above, exactly zero would have occurred if one or both parties were just carefully paying attention in the first place. If people are not incentivized to pay attention when the result of not paying attention could be serious injury or death, and the systems we are designing to try to save them requires people to pay attention, are we trying to engineer our way out of just plain laziness and stupidity?

IC

 

[thebard3]

...are we trying to engineer our way out of just plain laziness and stupidity?

That's mostly what I have to do here every day.

Brad Waybright

It's all okay as long as it's okay.

 

[IRstuff]

are we trying to engineer our way out of just plain laziness and stupidity

No, the reality is that it is nearly impossible to be 100% attentive 100% of the time; our attention wanders, we get tired, or we get distracted, even with just random objects within our field of vision. This is the promise of vision/AI systems; they maintain the same level of attentiveness and detection accuracy/efficacy whether it's one minute into the drive or 8 hrs into the drive. The big question is can we ever achieve the required detection accuracy without globs of false alarms? If the answer is no, then any sort of aiding might indeed be worse than none at all, since it lulls us into a false sense of security, making it that much harder to react/respond if something goes awry. Even when we're at our peak performance levels, our predator forebrains might be particularly lured to distractions that aren't actual threats, since our brains are not evolved for driving cars.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[DM61850]

Everyone would buck if someone suggested we lower the driving age to 12.

Hardly anyone says a thing when we let cars driven by computers on the road without any sort of screening even though a 12 year old would be a better driver.

 

[Lou Scannon]

I remain dubious that safety is the motivation for robotic vehicles. In my opinion, in the first place it is the elimination of expensive, paid human drivers. In the second place it is a broadening of the market, such that the ability to drive or afford a chauffeur is no longer a prerequisite for owning and operating a private motor vehicle. In the third place it is scheduled obsolescence, such that each generation of robotic vehicles is quickly made obsolete by lower cost and improved capability of the succeeding generation, thereby ensuring a sustained lucrative market.
Never mind that the gap between what they are working on now and what is needed to be barely viable in outdoor robotic driving is, in a word, incomprehensible.
The end game I see is that highly regimented zones are established to accommodate dumb robotic vehicles, in which human driven vehicles are banned, and outside of which purely robotic vehicles are likewise banned; or, alternatively perhaps, due to a few more spectacular episodes, the whole unfortunate experiment is written off as a failure and set aside for the foreseeable future.
Does anyone share my perception that this so called autonomous vehicle experiment is relying on the efforts of overconfident game programmers who are used to working within a confined, albeit large memory space, and do not comprehend concepts like hard real time, or infinity:

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[IRstuff]

In my opinion, in the first place it is the elimination of expensive, paid human drivers.

I think not, since that's the worst thing for a service industry, to have gigantic sunk costs. If you look at Uber's model now, they pay labor only, and have no liability for the hardware, or for insurance costs and repairs. Even assuming that the AIs can perfectly avoid inducing an accident, it just means that everyone around is a worse driver, so there will still be accidents, but now, you're talking about totaling a car that has $25k of additional hardware and the liability of any secondary incurred insurance risks. Uber had to settle with that pedestrian's family; had that been a human driver, Uber would have had washed their hands of the whole incident.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[human909]

I don't think ANYBODY really knows how and when the whole self driving cars will play out. It is sort of like asking computer programmers 40 years ago about having a bunch of apps and the internet in our pocket. AI has been imagined for well over half a century, but we don't really seem to be any closer.

https://en.wikipedia.org/wiki/HAL_9000

The end game I see is that highly regimented zones are established to accommodate dumb robotic vehicles, in which human driven vehicles are banned, and outside of which purely robotic vehicles are likewise banned;

The problem with that end game is that requires completely new infrastructrue and city layout. Which is no easy task. And IF you are going to do that then you might as well rethink the entire concept of a car. You might as well go to 1 person pods at 1/5th of the weight and without all the safety equipment or empty seats because the need for such things in a gated environment is removed.

I think not, since that's the worst thing for a service industry, to have gigantic sunk costs. If you look at Uber's model now, they pay labor only, and have no liability for the hardware, or for insurance costs and repairs. Even assuming that the AIs can perfectly avoid inducing an accident, it just means that everyone around is a worse driver, so there will still be accidents, but now, you're talking about totaling a car that has $25k of additional hardware and the liability of any secondary incurred insurance risks. Uber had to settle with that pedestrian's family; had that been a human driver, Uber would have had washed their hands of the whole incident.

Very good point.

But Uber continues to invest in this sector because they need to look like a TECH and not a taxi company with an easy to use app. Have a look at their financials and it is scary. They'll be bankrupt within the decade. Sure some TECH companies have spent years making losses, but that is because they were doing something nobody else could do AND/OR they had a captured market. Uber can't capture the market, it isn't a market that is a natural monopoly.

 

Uber have added literal killing to their long rap sheet of corporate shenanigans and legal abuses.
If any company can be described as toxic it is Uber.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

I remain dubious that safety is the motivation for robotic vehicles. In my opinion, in the first place it is the elimination of expensive, paid human drivers.

Its just another element of and a natural progression of the Amazon Project.
Translation: World Domination.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[IRstuff]

I still don't see the economics working out for driverless cars as a service

Assuming 22 hr of operation per day at $16/hr --> $128.5k revenue with 216.8 kmiles driven per year --> car replacement at least once per year.

Assuming 27 mi/hr average speed and $0.55/mi operating cost and $50,000 purchase price --> $169.3k per year amortized cost




TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[moon161]

But Uber continues to invest in this sector because they need to look like a TECH and not a taxi company with an easy to use app. Have a look at their financials and it is scary. They'll be bankrupt within the decade. Sure some TECH companies have spent years making losses, but that is because they were doing something nobody else could do AND/OR they had a captured market. Uber can't capture the market, it isn't a market that is a natural monopoly.

Yeah, I think cali passed a law recently with a number of test as to whether people are contractors or employees and it squarely places uber drivers as employees.

 

[dgallup]

If a car with driver can make economic sense and you can add the autonomous features for less than 50% extra then the autonomous car must be able to make a good business case. I think your $16/hr is way low. Probably could get 4 times that at least even without "surge pricing". Your average speed may be too high as well as your true cost per mile. These are going to be basic cars with little power driving on rock hard "eco" tires that never wear out. I envision plastic interiors that you can hose out, 3 cylinder engines, shit brown colors. Something about as nice as the inside of a subway car.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[SnTMan]

Something about as nice as the inside of a subway car.

Yeah. The trouble with public transportation is the, uh, public

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[hawkaz]

Johnny Cab?

 

[IRstuff]

If a car with driver can make economic sense and you can add the autonomous features for less than 50% extra then the autonomous car must be able to make a good business case.

But, it doesn't; the only reason it kind of works is that the car, insurance, etc., are sunk costs, and there's no attempt to recoup for that. The drivers are ignoring the economics and are only looking at the extra income, which is barely minimum wage.

27 mph is what my Prius reports as my average speed, over the course of a tankful and is probably 30% freeway, but it's Orange County, so there are brief instances of plausible freeway speeds, followed by minutes molasses-like progress.
$0.55/mi is roughly the GSA reimbursement rate for personal vehicles.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[hpaircraft]

Uber have added literal killing to their long rap sheet of corporate shenanigans and legal abuses.
If any company can be described as toxic it is Uber.

So now they're just straight up equipping each car with a bone saw?