Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash • Hacker News Board

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#441

Post by _ea1k » Thu, May 02, 2019, 1:23 PM UTC

radcon wrote: post #436

helloindia wrote: post #382

A more nuanced dissection of whose fault it was or could be[0]. "Huang was apparently fooled many times by Autopilot. In fact, he reportedly experienced the exact same system failure that led to his fatal crash at the same location on at least seven occasions." ........... "Huang knew that Autopilot could not be relied on in the circumstances where he was commuting along the 101 freeway in Mountainview, California. Y…

> Yet he persisted in both using the system and ignoring the alerts that the system apparently gave him to put his hands on the wheel and take full control. "Hands not detected" does NOT mean "hands not on wheel". Tesla relies on driver input (torque) to "detect" hands, which means you could have both hands on the wheel but sill get the warning to place your hands on the wheel. It seems like most people are assuming…

I agree, but I'm also not sure that it is a particularly important distinction. He was repeatedly warned that attention is required with this system and yet clearly was not paying attention at the time of the crash.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#442

Post by hvidgaard » Thu, May 02, 2019, 1:26 PM UTC

hnlmorg wrote: post #426

hvidgaard wrote: post #384

Earlier quoted context omitted.

> I've travelled by road a fair bit around different countries and I assure you that a lot of high volume maintained roads still fall into the categories I outlined. Moreover, a lot of high volume roads aren't maintained. I honestly don't think you would even get remotely close to covering 80% of the roads for 20% of the cost. I did not say 80% of the road, but 80% of the travel. Most of the kms driven are on highway…

> I did not say 80% of the road, but 80% of the travel. Most of the kms driven are on highways, ring roads, and main roads into larger cities. I don't think your experiences are typical for most people the world over. For example in the UK, we don't have that many motorways so most of the commutes by cars are along busy "normal" roads or dual carriageways (which are a little like motorways but with frequent junctions…

I'm from Denmark, and my experience ranging from all the Nordic countries, Germany, Spain, France, Italy, and a bit in other countries but not enough to have any real experience.

We could turn this system around. It would reliably cover the vast majority of semi transport if it just worked on major highways in Europe. Drive the truck to the highway on ramp, jump out and let it drive 1000km before someone picks it up and drive it "the last kilometer".

I think your timeline is overly pessimistic. 55 years ago way 1964 - a lot have happened in that time frame, and we put a man on the moon and a rover on mars in far less time. It can happen significantly faster. If we decided to start developing now, I believe that within 5-10 years we would get near a specification and legislation. That leaves changing the cars and that only have to be done gradually. We know so little about this problem domain that it's best done gradually and carefully.

I think we can agree to disagree.

I'm not convinced that weak AI will be anything but an enhancement to the human driver.

It's my belief that we don't even have a grasp of the AI domain to do this yet. Perhaps we will get it, perhaps we will not. Nothing that has happened in the AI space is new from the theoretical point of view. Neural networks was discovered in 1943, we just recently got the computational power to use it. But it doesn't scale and it's a black box approach. And it's easily fooled (see recent reports of a simply sticker on the road making a Tesla drive against traffic in the wrong lane).

The strong AI field haven't seen any real progress or breakthrough in a long time, and people still argue whether it's even possible. To put it into numbers, a worm, which we have not successfully modeled the brain of yet, have around 300 neurons. A human have 10^11, each witch have a far a greater connective complexity . Strong AI is not coming in the next 100 years, if at all.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#443

Post by _ea1k » Thu, May 02, 2019, 1:28 PM UTC

studentrob wrote: post #420

strainer wrote: post #409

Earlier quoted context omitted.

You are claiming that the accuracy of Telsas language is outrageously deceptive, using an outrageously made up example of deception: "six seconds prior to the collision = It is any six seconds before the crash" "This is the kind of sketchiness that makes people wary of Tesla." Its the kind of sketchiness that make me wary of the stuff people post about Tesla. If the driver had multiple warnings - about anything, any…

According to Tesla, > the only way for this accident to have occurred is if Mr. Huang was not paying attention to the road [1] Tesla has made a habit of drawing conclusions about investigations that have just begun. It is sketchy, and people have noticed. [1] https://www.businessinsider.com/tesla-says-fatal-model-x-aut...

I don't understand. What other theory are you proposing than that he wasn't paying attention?

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#444

Post by cptskippy » Thu, May 02, 2019, 1:32 PM UTC

CoolGuySteve wrote: post #171

cptskippy wrote: post #111

Earlier quoted context omitted.

AAA tests found 40% of AEB systems failed to stop a collision at 30mph and 60% at 45mph. Granted they're not causing them but those numbers still aren't great and get worse the faster you're traveling. https://www.computerworld.com/article/3111407/aaa-automatic-...

At some speed stopping the collision becomes difficult/impossible. But in the meantime I'd vastly prefer colliding after significant breaking than not.

Absolutely, and most AEB systems can't or don't do that. Everyone wants a safe AEB that functions correctly in all situations but we're just not there yet.

https://arstechnica.com/cars/2018/06/why-emergency-braking-s...

I think the real question is did Tesla mislead customers into believing their system was more capable than it actually was, thus resulting in them relying on a system when they shouldn't have?

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#445

Post by _ea1k » Thu, May 02, 2019, 1:40 PM UTC

bufferoverflow wrote: post #433

jfk13 wrote: post #413

Earlier quoted context omitted.

But it's perhaps not unfair to sue them for misleading advertising that leads to people getting killed. "The person in the driver's seat is only there for legal reasons. He is not doing anything. The car is driving itself." creates very different expectations from "While the car is driving itself, the person in the driver's seat is keeping his hands on the steering wheel and his attention on the road at all times."

Doesn't Tesla push the second scenario at the moment?

They literally tell you to put your hands on the wheel and pay attention every time you don't, and sometimes even when you do. So yes, they push the idea that awareness is required.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#446

Post by FartyMcFarter » Thu, May 02, 2019, 1:42 PM UTC

tim333 wrote: post #373

FartyMcFarter wrote: post #369

Earlier quoted context omitted.

> They specifically tell you it is NOT a fully autonomous autopilot and that's it's more of an advanced cruise control. Not here they don't: https://www.tesla.com/en_GB/autopilot?redirect=no Look at the caption in the beginning of the first video. They say the driver is only there for legal reasons and not doing anything.

"Current Autopilot features require active driver supervision and do not make the vehicle autonomous." From your link.

Tesla needs to get their story straight. They're trying to talk out of both sides out of their mouth, and using whichever statement is convenient at different places.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#447

Post by radcon » Thu, May 02, 2019, 1:44 PM UTC

_ea1k wrote: post #441

radcon wrote: post #436

Earlier quoted context omitted.

> Yet he persisted in both using the system and ignoring the alerts that the system apparently gave him to put his hands on the wheel and take full control. "Hands not detected" does NOT mean "hands not on wheel". Tesla relies on driver input (torque) to "detect" hands, which means you could have both hands on the wheel but sill get the warning to place your hands on the wheel. It seems like most people are assuming…

I agree, but I'm also not sure that it is a particularly important distinction. He was repeatedly warned that attention is required with this system and yet clearly was not paying attention at the time of the crash.

> clearly was not paying attention at the time of the crash.

This is the fundamental point that I'm disagreeing with though. As far as I know, Tesla has a very crude system for determining driver attentiveness. They rely ONLY on steering wheel torque, which is an extremely unreliable indicator. They don't do any sort of head or eye tracking. So when Tesla claims a driver wasn't paying attention, all it really means is "we're not sure if their hands were on the wheel".

My car uses the same method, and I constantly get warnings to place my hands on the wheel even when they're both already on the wheel. It's possible to have your hands on the wheel without applying any torque, and that's where this system fails.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#448

Post by kurtisc » Thu, May 02, 2019, 1:48 PM UTC

solveit wrote: post #306

areoform wrote: post #96

I would like to add over here that Tesla does indeed do sensor fusion in their cars. Their Autopilot combines radar and ultrasound with vision to decide where to drive. Commentators bringing up LIDAR are jumping the gun by assuming that this scenario isn't something that these sensors wouldn't have detected in either combination or individually to spot the anomaly. The problem over here is likely to be software due t…

> What if the neural net or the system used to detect obstacles didn't see it because the precise configuration of the data fooled it? And if that's the case then what's next? How do we decide when it is okay for safety-critical systems to be opaque? How do we deal with autonomous driving if the conclusion comes out to be a "no" for this case? How should broader society deal with a yes? And who decides all of this in…

People are very concerned about unsafe human driving and the liability for the massive number of deaths and damages it causes. It's the reason we have insurance and a big reason why we have self-driving aspirations at all.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#449

Post by bloak » Thu, May 02, 2019, 1:55 PM UTC

marmshallow wrote: post #227

lwansbrough wrote: post #182

Earlier quoted context omitted.

Telsa's bet, and I think it's a really fair one, is that if humans only need two eyes to navigate as well as we do, a car with >2 eyes could eventually be as good or better than humans. On the other hand, humans do not have lidar systems in their head, and so lidar is the real unproven technology. It's also clear that lidar is especially lacking in situations which require seeing longer distances (like say, driving!)…

Sure, humans can drive with two eyes and a brain. But do Tesla's cameras have the amazing abilities of the human eye, and do his "FSD" chips have the power of the human brain? We'll find out in a few years...

Humans can even drive with one eye and a brain.

Re: Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

#450

Post by bloak » Thu, May 02, 2019, 1:56 PM UTC

CoolGuySteve wrote: post #171

cptskippy wrote: post #111

Earlier quoted context omitted.

AAA tests found 40% of AEB systems failed to stop a collision at 30mph and 60% at 45mph. Granted they're not causing them but those numbers still aren't great and get worse the faster you're traveling. https://www.computerworld.com/article/3111407/aaa-automatic-...

At some speed stopping the collision becomes difficult/impossible. But in the meantime I'd vastly prefer colliding after significant breaking than not.

And colliding after significant braking would be even better.