2 dead in Tesla accident "Noone wasdrivingthe car" 15

[MartinLe]

https://www.click2houston.com/news/...er-fiery-tesla-crash-in-spring-officials-say/

“no one was driving” the fully-electric 2019 Tesla when the accident happened. There was a person in the passenger seat of the front of the car and in the rear passenger seat of the car.

the vehicle was traveling at a high speed when it failed to negotiate a cul-de-sac turn, ran off the road and hit the tree.

The brother-in-law of one of the victims said relatives watched the car burn for four hours as authorities tried to tap out the flames.

Authorities said they used 32,000 gallons of water to extinguish the flames because the vehicle’s batteries kept reigniting. At one point, Herman said, deputies had to call Tesla to ask them how to put out the fire in the battery.

 

[IRstuff]

That’s how Musk operates. He’s on the cutting edge.

I think you are buying into Musk's hype, or you don't have the sarcasm flag turned on. The Tesla's lane following algo is ludicrously lame; it doesn't even care that the lane it's following is following the map the navigation system knows it's following, nor that none of the other cars are going in that direction. Musk's obsession with only using cameras for navigation resulted in the algo thinking that the side of a semi was a high sign that it could go under and ignored the fact that the semi turned in front of the Tesla and then just magically disappeared. Musk and Uber employ too many image processing scientists and not enough systems engineers.

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

 

[LionelHutz]

You beat me to it.

 

[SwinnyGG]

That’s how Musk operates. He’s on the cutting edge.

Designing a system by deliberately ignoring the most robust technology available in favor of what is easier to procure and less expensive to develop, at the expense of ultimate safety and ultimate capability, is about as far from 'on the cutting edge' as you can get.

That's exactly what Tesla has done, at Musk's explicit direction. The cult of personality that has developed around Musk's perceived genius is myopic beyond belief.

 

[BrianPetersen]

Elon Musk strikes me as someone who will not take no for an answer. If someone tells him no, they're out and replaced with someone who says yes. This is what you end up with.

 

[itsmoked]

I think you've got the wrong term in use. It's not "cutting edge" it's "bleeding edge". Unfortunately it's literal in these cases.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Tomfh]

I think you are buying into Musk's hype, or you don't have the sarcasm flag turned on. The Tesla's lane following algo is ludicrously lame; it doesn't even care that the lane it's following is following the map the navigation system knows it's following, nor that none of the other cars are going in that direction.

I've driven a Tesla that tried to swerve offroad and down an embankment at 110kmh, so I'm quite aware of the limitations of his methods! Nonetheless he is pushing the technology. Maybe his camera approach will fail. Who really knows. We humans do ok with two limited cameras (sometimes only one), and so its not inconceivable that Musk's approach will pay off. He has a lot of hype surrounding him, but it's not all smoke and mirrors.

 

[LionelHutz]

The fact that Teslas which were supposed to be fully self driving by now are still just following the lane lines and the fact that Teslas continue to drive into solid objects both give big hints about how well the camera only approach is working.

I'm not sure how much he ultimately cares anymore. Going to space seems to be his new pet project.

 

[davefitz]

Agree that his main point of interest is space- it will be far more profitable than EV's given the large amount of new competition now being offered by european and japanese carmakers.

"...when logic, and proportion, have fallen, sloppy dead..." Grace Slick

 

[IRstuff]

We humans do ok with two limited cameras (sometimes only one)

We have brains with millenia of learned visual processing tricks AND more importantly we have memory and "lessons" learned capacity. Tesla is more of a guess and check approach.

One issue with hiring only image processing whizzes is that they typically don't start off with a radar tracking problem analogy, so each image is a standalone thing to process; this was the case with Uber's experiment that caused the death of a pedestrian, where 400 ft prior to the collision, the pedestrian was detected, and clearly moving on a collision course, but the code didn't have squat to deal with that.

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

 

[drawoh]

IRstuff,

The code was not connected to the controls and the brakes. The pedestrian was detected in good time.

--
JHG

 

[LionelHutz]

The Uber code was driving the car, so it certainly was connected.

 

[IRstuff]

The code was not connected to the controls and the brakes.

It was indeed connected, since it was in charge of both lane keeping and collision avoidance; had there been a detectable car already in the lane, the Uber would have most certainly braked and warned the "safety" driver in plenty of time.

The pedestrian was detected in good time.

No, that is not correct; the safety driver was warned less than 0.6 seconds before impact. The algos detected "something" much earlier, and what messed up the software was target ID kept changing, but there was no fail-safe algorithm that simply kept track of "any" object that might interfere with the safety, and there was apparently nothing in the software that attempted to do the basic radar tracking function of collision course detection, which would have said something like, "I don't know what I found, but it's on a collision course with us." THAT, it could have done within the second detection, which was still a few seconds from impact. And all this was done with a LIDAR, so even fewer excuses possible. But, Tesla's algorithms operate similarly, as the collision of a Tesla and a tractor-trailer demonstrated; it detected a truck turning in front of it, and then decided the truck was a highway sign, which was likely moving, too.

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

 

[dik]

This just ended up in my in basket...

https://www.automationmag.com/auto-...um=email&utm_campaign=&*URLENCODE({{*JobID}})

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

 

[IRstuff]

Tesla's latest 10-Q filing concedes that they might never achieve what Musk continues to promise

“We are developing self-driving and driver assist technologies to rely on vision-based sensors, unlike alternative technologies in development that additionally require other redundant sensors. There is no guarantee that any incremental changes in the specific equipment we deploy in our vehicles over time will not result in initial functional disparities from prior iterations or will perform as expected in the timeframe we anticipate, or at all.”

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

 

[davefitz]

the question never asked is " why the merry Eff do we need self driving cars?" Where is the demand?

"...when logic, and proportion, have fallen, sloppy dead..." Grace Slick

 

[MacGyverS2000]

People could get a LOT done if they didn't have to concentrate on the drive itself... I know I accomplish a lot of work on flights, barring the time wasted going through security, standing in lines, driving to/from, etc. (which is a large portion).

Dan - Owner

http://www.Hi-TecDesigns.com

 

[JohnRBaker]

It's not so much having a car where the driver can do something else while 'driving', it's to get to where the technology will allow us to have fully autonomous vehicles, sort of like a driver-less Uber or long-haul robotic trucks.

John R. Baker, P.E. (ret)
EX-'Product Evangelist'
Irvine, CA
Siemens PLM:
UG/NX Museum:

The secret of life is not finding someone to live with
It's finding someone you can't live without

 

[JStephen]

With a reliable self-driving car, you could nap, read, watch TV or work during your commute.
In the event you're getting too old or too blind to safely drive, it'd be very handy as well.
The biggest problem I see is that unless there is a "Drive like a lunatic" setting, a goodly percentage of current drivers wouldn't ever want one. It'd be as popular as the 55 mph speed limit with them.

 

[SnTMan]

"Drive like a lunatic" setting

Hey! Time is money!

Also I don't see self-driving / autonomous vehicles being tolerated by Ultra Type A's, either.

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

 

[Spartan5]

I don’t think the vision for autonomous cars involves a lot of individual sales.

I think the vision is to greatly reduce car ownership on an individual basis. The transportation paradigm because ones of autonomous ride sharing. With cars only stopping when they need to charge. Load balanced by surge pricing and/or contractual agreements at negotiated rates for a ride at a particular time of day. Pay a premium, and maybe the rest of the automated cars will clear a path for you. It turns everything into algorithmasic “just in time” optimization problem to be solved.

Hey, Alexa, I need to go to go to the grocery store. But that goes away too. Because all of these vehicles will have side-hustles delivering packages/food/groceries in their spare time.

The vehicles become the network, instead of the roads. It’s all the Uber model; sans those pesky driver “contractors” and having to treat them like humans.