Tesla issues strongest statement yet blaming driver for deadly crash • Hacker News Board

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#801

Post by nojvek » Thu, Apr 12, 2018, 5:23 PM UTC

maxxxxx wrote: post #554

Animats wrote: post #552

Tesla's system demands more from drivers than manual driving. The driver has to detect automation failures after they occur, but before a crash. This requires faster reaction time than merely avoiding obstacles. Here's a good example - a Tesla on autopilot crashing into a temporary road barrier which required a lane change.[1] This is a view from the dashcam of the vehicle behind the Tesla. At 00:21, things look norm…

" Tesla's system demands more from drivers than manual driving. The driver has to detect automation failures after they occur, but before a crash. This requires faster reaction time than merely avoiding obstacles." I have been thinking that too. At what point do you decide that the autopilot is making a mistake and take over? That's an almost impossible task to perform within the available time.

If it’s a car that doesn’t require human driver to have her/his hands on the steering wheel at all times, and it crashes. It’s the car’s fault. i.e it collided into another semi-stationary object when it could steer away or brake.

This should be a law.

#1 job of any self driving system is to avoid obstacles. If you can’t do that reliably, it’s a fail!

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#802

Post by nojvek » Thu, Apr 12, 2018, 5:27 PM UTC

maxxxxx wrote: post #757

omarforgotpwd wrote: post #593

Earlier quoted context omitted.

Try it. it’s not impossible. I do it all the time. Have you ever noticed how passengers are still paying attention to what’s going on and sometimes even turn their head to see where the car is turning? it’s like that, except you’re in the drivers seat. it’s not more work than driving.

When I am a passenger I do not always pay attention, only sometimes. And if something happens it will the driver's fault, not mine. It would be a very stressful ride if I had to always observe the driver and if I didn't intervene anytime every problem would be my responsibility.

I definitely think auto-pilot is a really bad name. It should be “advanced cruise control”. Because essentially that’s what it is. Follow the lane and maintain distance behind a car.

Also unless certified on certain roads with clear markers, drivers should be required to have hands on wheels at all times.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#803

Post by rohit2412 » Thu, Apr 12, 2018, 5:32 PM UTC

akvadrako wrote: post #736

rohit2412 wrote: post #170

Earlier quoted context omitted.

I think the solution is to hold Elon Musk/Tesla liable for their misleading statements on FSD, Tesla Network, AP2 and their timelines about future capability. Future promises if made should come with renaming the product to highlight its capabilities.

That's potentially very complicated. As long as drivers are required to purchase insurance and insurance is required to pay out fair sums for accidental death, the accountability should come automatically in the form of higher rates.

Do you also hold the same view regarding medicines? That the FDA shouldn't regulate medicines, and manufacturers can make insane promises? After all, we love free market over human lives.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#804

Post by astrodust » Thu, Apr 12, 2018, 5:46 PM UTC

semi-extrinsic wrote: post #728

michaelt wrote: post #725

Earlier quoted context omitted.

Seems to me in the video there's no intervention needed at the 21 second mark, and the car's already hit the barrier at the 22 second mark. Youtube doesn't provide sub-second time marks, but I'd estimate it's 21.5 second before I'd expect the car to start turning, giving the driver 0.5 seconds to notice the failure, take control of the vehicle, and execute the swerve required. The Queensland government estimates it t…

> giving the driver 0.5 seconds to notice the failure, take control of the vehicle, and execute the swerve required. If you have to be constantly at this level of alertness, what is even the point of having an autopilot?

This is the automation paradox, and one that Tesla itself has been facing with their assembly lines.

If you over-automate something you end up having to spend more time supervising and fixing it than if you'd just done the thing yourself in the first place.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#805

Post by danso » Thu, Apr 12, 2018, 5:59 PM UTC

sokoloff wrote: post #702

danso wrote: post #397

Earlier quoted context omitted.

> You are reading into that text more than you should. The page's title is "Autopilot | Tesla". It is the first result for "tesla autopilot" in search results. And "autopilot" appears 9 times on the page. So if that's not an intentional attempt to mislead consumers into conflating Autopilot with "full self-driving", then what would such an attempt look like, hypothetically?

Hypothetically, it might say that all Tesla cars being built today have both the hardware and software to make full self-driving possible. It doesn’t; what it does say is perfectly clear to me.

That would ostensibly be a lie.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#806

Post by sokoloff » Thu, Apr 12, 2018, 6:02 PM UTC

danso wrote: post #805

sokoloff wrote: post #702

Earlier quoted context omitted.

Hypothetically, it might say that all Tesla cars being built today have both the hardware and software to make full self-driving possible. It doesn’t; what it does say is perfectly clear to me.

That would ostensibly be a lie.

Yes, I know.

It remains an answer to "what would such an attempt [to intentionally mislead consumers] look like, hypothetically?"

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#807

Post by eh78ssxv2f » Thu, Apr 12, 2018, 6:18 PM UTC

product50 wrote: post #704

eh78ssxv2f wrote: post #691

Earlier quoted context omitted.

Sorry, comparison with airplanes is not fair in my opinion. The purpose of the autopilot is to increase safety for drivers that are not attentive. However, it turns out that Tesla requires drivers to be even more attentive than regular vehicles. This is not a sign of progression.

Not true in my opinion. While for now the positioning of autopilot is to increase safety for non attentive drivers, eventually autopilot's main goal is to have completely self driven cars and that is where the industry is heading (and Tesla wants to be in the forefronts of it). We can differ in opinions but I strongly feel that the eventual utopia of self driving cars will be much safer than the current world. And we…

> While for now the positioning of autopilot is to increase safety for non attentive drivers, eventually autopilot's main goal is to have completely self driven cars and that is where the industry is heading (and Tesla wants to be in the forefronts of it).

In that case they should issue a recall, and not use real users as their test drivers. Once they get to their eventual goal, they can resume selling autopilot cars.

> We can differ in opinions but I strongly feel that the eventual utopia of self driving cars will be much safer than the current world. And we are making progress on this daily. There will be few unfortunate incidents but in the long term a lot more lives will be saved.

I totally agree with you that in the long term self-driving cars are much safer. What I do not agree with is Tesla's approach towards achieving that goal. They are selling an unsafe product while marketing it as much safer than the existing products. This is just plain fraud.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#808

Post by Dylan16807 » Thu, Apr 12, 2018, 6:55 PM UTC

greglindahl wrote: post #628

Dylan16807 wrote: post #537

Earlier quoted context omitted.

Steering toward a barrier is not fine for any level.

Er, no, that’s not how it works. Level 2 is not expected to be perfect in that fashion. You want level 3. Automatic cruise control is also not expected to be perfect at slamming on the brakes to avoid rear-enders. The driver is responsible.

There is a huge difference between not-avoiding an accident and causing an accident. It didn't just drift, it actively pointed itself toward the barrier. That shouldn't happen at level 2, or level 3, or level 0.

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#809

Post by paulryanrogers » Thu, Apr 12, 2018, 6:57 PM UTC

Double_a_92 wrote: post #700

paulryanrogers wrote: post #589

Earlier quoted context omitted.

There seems to be a valley between the way the user expects something works and how it actually does. When users detect a difference they may compensate, if they survive long enough to learn. Of course if users get to used to such differences then those can become 'features' which only makes progress harder.

The problem is that you don't know what the car is going to do. E.g. if you see a barrier you would mentally prepare to steer. But if you trust the car you don't prepare and then suddenly have to take over at the last moment when you realize the car is doing something wrong. That means the autopilot doesn't help at all since you would have to basically drive manually all the time anyway.

In this case I'd agree that the cons outweigh the pros. Surprising Tesla can legally market the feature or even offer it to the general public in its current form.

EDIT: typo

Re: Tesla issues strongest statement yet blaming driver for deadly crash

#810

Post by Eruditass » Thu, Apr 12, 2018, 7:48 PM UTC

salawat wrote: post #770

corney91 wrote: post #742

Earlier quoted context omitted.

> If they had the ability to measure confidence in the system, they would use it to issue warnings or disable autopilot Yeah, that's basically what I'm asking for, maybe a warning light at a certain threshold would be a better default, personally I'd still find a number more trustworthy though. > The "confidence meter" would almost certainly be 100% right up until it crashes into an obvious obstacle If image recognit…

The problem is, given a neural network, the decided course of action IS at 100% confidence. Unless you design the network to specifically answer "how similar is this?" and pair it with a training set and the output of another executing neural net, a "confidence meter" isn't possible. This is one of the trickier bits to wrap one's mind around with neural networks. The systems we train for specialized tasks have no con…

Have you actually worked on applying neural networks in the real world?

Calibrating the probabilities of machine learning algorithms is an old problem. By nature of disciminative algorithms and increasing model capacities, yes training typically pushes outputs to one extreme. There is a ton of information still maintained which can be properly calibrated for downstream ingestion, which anyone actually trying to integrate these into actual applications should be doing.

Most recently:

https://arxiv.org/abs/1706.04599 https://arxiv.org/abs/1802.03916