Autonomous Vehicles NOT Ready for Prime-Time

[LPSISRL]

Kuro, I'm going to nit-pick here...

"The following is from National Highway Traffic Safety Administration (NHTSA) data:
In 2013, there were 32,893 deaths from motor vehicle accidents in the United States.
In 2016, there were 37,461 deaths from motor vehicle accidents in the United States.
That's an increase in over 4,500 people dying despite only a difference of 3 years. 3 years isn't much time for automotive technology to improve much, but it sure had a difference on average fuel prices in the nation:
In 2013, the average price for Regular gasoline in the US was $3.51.
In 2016, the average price for Regular gasoline in the US was $2.14.
The lower cost of gasoline had more people driving in 2016 than in 2013. This then resulted in more accidents and more people dying."

I think you're missing at least one or two very important statistics to connect lower fuel prices with casualties. 1. That the lower fuel prices put more cars on the road. You're making an assumption without the facts to back it up. 2. With cheaper fuel prices, cars on the road were driven more miles meaning that they were on the road longer which would increase the odds of being in an accident. While both are logical and probably true, without the facts to back them up they can't be used to prove a point.

[LPSISRL]

OK, I've looked at this video quite a few times and stopped it at the point when the woman's feet first come into view. It's like what, 50-75 feet down the road? If my estimate is way off, how far is she? I've upgraded my headlights to HIDs and on a dark road, I'm sure I'd have picked her up a couple hundred feet away. On several occasions now when driving on dark roads even at 50-60 MPH I estimate when I would pick something up with my headlights and I firmly believe that my car would have picked her up in plenty of time to stop. She didn't get picked up when she stepped off the curb, she wasn't picked up by the Volvo's headlights until she was a couple of feet into the lane it was driving in. The video is at 3 seconds when her feet are first visible and at 4 seconds at impact. At 38 MPH, 1 second is 55 feet and 2 seconds is 110. So she was somewhere in between.

Am I missing something here? Is it just the video or was there something not right with the Volvo's lighting system? Driving lights only? I see a lot of idiots driving around after dark with only their DRLs but it really seems to me that she should have been picked up in the Volvo's headlights much further away.

[Ronzuki]

Comments from last week’s Uber autonomous car crash…

“For a successful technology, reality must take precedence over public relations, for nature cannot be fooled.”

“What is the point of these driverless vehicles? We can call all learn to drive & pass a test. This technology is being done because we can, not because it’s useful. A system once set up would be the perfect target for terrorist organizations, it may take them years to hack in, but you would only need to do this once to cause massive problems & or death.
Driver assistance is a great way forward, but the driverless car is good money wasted. Spend it making the roads nice to drive on & safe.”

“This is the latest in a long line of accidents involving autonomous vehicles, and the second fatality. Also there are reported many examples of speeding and jumping red lights. The current state of the technology is not developed enough to be allowed on public roads. Any other technology that has killed two people in such a short space of time would be recalled and revised. Current testing in cities is putting the public at risk and should be stopped instantly. I hope our government rethink their push to get these vehicles on our roads, but this is doubtful.”

https://www.theengineer.co.uk/uber-cras ... 6f52ee9238

[Ronzuki]

What the hell are you people looking in at? Must be some other video...
The AGV didn't see the object, at all, until it ran it over...it didn't slow, it didn't jam on the brakes (nose dive), it didn't attempt to swerve...IT FLAT OUT DIDN'T SEE THE OBJECT IN ITS PATH. Now pay attention...this is important in automation...it's path is anywhere it can travel now, not just where it is travelling. Try and grasp that concept while commenting.

Yes Darwinism prevailed in this instance, no doubt. Not my primary point, or argument, in this thread, hence the thread's title.

Suppose that were a little kid that didn't know any better? Would it have made any difference in the outcome? Not a damn bit. An AGV would certainly not give a crap, tree branch, car part, box, animal, human, whatever. The technology FAILED EPICALLY. Keep on twisting and massaging the facts, not buying into any of it. If your going to put tech as this out in public, it better damn well be more attentive than what it's attempting to replace.

[LPSISRL]

I thought this article beneficial for any who would like an overview of the systems that AVGs use.
https://techcrunch.com/2018/03/19/heres ... destrians/

[Ronzuki]

Lastly, regarding this: "Again, the PUBLIC roadways are clearly not place to carry out mission critical testing. In my world, that's just plain old common sense. But apparently common sense truly has met its end, is dead and seemingly buried."[/i]
It would seem like common sense on a superficial level until you really think about it critically. What is automated driving supposed to do? It's supposed to drive on public roads. How do you even create an environment to mimic the complexities of public roads if they aren't just that? It's logically flawed. The best example would be from the following scenario: an automated vehicle company never tests on public roads. It always runs simulations, testing, and experiments in a laboratory setting, closed course, or computer software simulations. It then releases a product for the market after years of testing. A short while after market release, it causes an accident. It is then determined it was never tested and vetted on public roads though it was designed to run on them. Guess what people would then complain? "Why wasn't this tested on the very roads it was designed to run on before being released to the public!?"


We tested AGVs in a 40,000 square foot building with alot of obstacles and scenarios to challenge the AGVs before they were put in to service in a customer's plant where they could damage equipment and shutdown production. Hollywood can create fake reality, it would stand to reason ya might want to do that first with a driverless weapon before letting it roll down the public roads.

See, you can only "fully vet" something after testing it in the very environment it was designed to operate on. Furthermore, Uber was doing just that. Uber isn't an automation company.

Oh yeah, that's abundantly clear, and, a huge part of the problem.

http://www.dailymail.co.uk/news/article ... felon.html

They are a ridesharing company. They are actually testing automation for the purpose of their main business of driving people around. And make no mistake, putting more people out of work...the main focus of modern day automation. Hence, they were in the process of testing and vetting the technology. This is why there was a USELESS human back-up driver getting paid to joy-ride in the cool Uber-mobile when they should have had an engineer in there observing what the AGV wasn't seeing and none of the automated Ubers are driving actual Uber customers around.

Now, it's come apparent that some people will never feel safe about automated driving until it's 100% free from flaws and is absolutely perfect. That's just a naive perspective when uncontrolled variables are constantly involved. One cannot be 100% flawless unless the environment it is in is that way too. This recent Uber crash is an example of how that's an unrealistic expectation. A woman caused the accident by jaywalking at night from the dark onto the path of a moving vehicle. Automated or not, it would have resulted in a collision in which the woman would have likely died as she did. I simply think people have unrealistic expectations from automation technology that they would never apply to themselves. Realistically, automation can't be absolutely perfect. It just needs to be better than a human driver. This incident doesn't prove automation is flawed in ways a human driver isn't.

You've just illustrated many of the reasons this technology should have never been permitted in the first place. Thank you.

[LPSISRL]

I previously posted about the headlight illumination for a couple of reasons. I'm interested in knowing the technical reason(s) for the failure and I'm also interested in the debate on whether we should have self-driving cars at all. The questions I'm looking for answers to are why the radar and lidar did not pick up the woman crossing the road before she visibly came into view illuminated by the headlights. If they were functioning as designed, the Volvo should have braked hard, veered left or both. But it did nothing until after impact as I understand it. We know the woman crossed from the center median going to the right across a full lane of roadway before proceeding into the Volvo's path. This was not a deer bounding from the side of the road entering suddenly into the path of the vehicle. This was a person walking no more than 3 or 4 mph.

I am also interested in knowing why the Volvo got so close before the headlights illuminated her. Doesn't it seem odd to anyone else how close she was before coming into view? Or am I wrong there?

This debate reminds me of Jurassic park where Dr. Ian Malcom says, "Your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should." Replace "scientists" with "engineers". Both software and hardware in this case.

Coincidentally, I watched "I Robot" last night. Scary if you replace the robots with self-driving cars.

[Ronzuki]

hmmm...how 'dark' was it really? Always thought it appeared far too dark.

At least one person wanted to show the Uber video to be flawed regarding how dark it was. The actual intersection appears at mark 0:32…

Any human would have been able to see, and at bare minimum, take evasive action if not stop completely. Any human focusing on their driving that is.

So who's dash-cam (again more manipulateable tech) is more accurate and who's is full of the proverbial chit? My money's on the non-Uber owned cam being a little more realistic for the environment.

Any members in the area? Take a night-time ride and let us know your thoughts.

[KuroNekko]

I think you're missing at least one or two very important statistics to connect lower fuel prices with casualties. 1. That the lower fuel prices put more cars on the road. You're making an assumption without the facts to back it up. 2. With cheaper fuel prices, cars on the road were driven more miles meaning that they were on the road longer which would increase the odds of being in an accident. While both are logical and probably true, without the facts to back them up they can't be used to prove a point.

Think about it and it becomes harder trying to disprove this correlation rather than proving it. How else can lower fuel prices be correlated with increased motor vehicle accident deaths? No other variable could have changed so much in 3 years other than fuel prices.
But here's some evidence anyway because I will admit it's an assumption (though a correct one):
According to the NHTSA:
In 2013, the total number of miles driven in the US was estimated at under 3 trillion.
In 2015, the total vehicle miles traveled increased to 3.15 trillion.

Hence, it's very evident that cheap fuel prices greatly contributed in more people driving and for longer. This then increased the risk of accidents and therefore, more people died on the road. It simply follows logic. Driving is a dangerous activity and traffic accidents are a leading cause of death for many. The more of it you do, the higher the chances are that you'll end up dying from it.

[LPSISRL]

I think you're missing at least one or two very important statistics to connect lower fuel prices with casualties. 1. That the lower fuel prices put more cars on the road. You're making an assumption without the facts to back it up. 2. With cheaper fuel prices, cars on the road were driven more miles meaning that they were on the road longer which would increase the odds of being in an accident. While both are logical and probably true, without the facts to back them up they can't be used to prove a point.

Think about it and it becomes harder trying to disprove this correlation rather than proving it. How else can lower fuel prices be correlated with increased motor vehicle accident deaths? No other variable could have changed so much in 3 years other than fuel prices.
But here's some evidence anyway because I will admit it's an assumption (though a correct one):
According to the NHTSA:
In 2013, the total number of miles driven in the US was estimated at under 3 trillion.
In 2015, the total vehicle miles traveled increased to 3.15 trillion.

Hence, it's very evident that cheap fuel prices greatly contributed in more people driving and for longer. This then increased the risk of accidents and therefore, more people died on the road. It simply follows logic. Driving is a dangerous activity and traffic accidents are a leading cause of death for many. The more of it you do, the higher the chances are that you'll end up dying from it.

That's exactly the piece of data left out. 150 million more miles driven in 2015 than in 2013. Sounds like a lot and it supports your hypothesis. But maybe not. In 2015 there were 7.8 million more cars registered than in 2013. (263.6 vs 255.8) Assuming, (making an ass of you and me) 12K miles on the average per year for those additional cars, that's 93.6 million additional miles driven simply by the increase in the number of cars. Now you spread the real difference of 56.4 million miles between the other 255.8 million cars and the increase does not look very significant. That's .2 miles per car. What if the average number of miles driven was 15K or 18K? At 18K per year, the number of additional miles driven is about totally wiped out by the additional cars. BUT... How many more drivers were there? How many were actually on the roads and not sitting in driveways and dealer lots? That's the problem with statistics.