After 6 months of working fine, Tesla software update drives at barriers again • Hacker News Board

Re: After 6 months of working fine, Tesla software update drives at barriers again

#251

Post by rayiner » Thu, Mar 21, 2019, 2:01 AM UTC

0815test wrote: post #210

quickthrower2 wrote: post #186

Earlier quoted context omitted.

Cruise control is useful to me because it means I won’t accidentally speed. But having to hold the wheel means I need to pay attention. If the car is driving itself human nature means you might not pay attention. Also if the car does something weird you have to decide in a split second if it is because of a hazard you haven’t seen but the car has, or because the software went wrong.

You do have to hold the wheel, precisely because you might find yourself having to assess what the car is doing "in a split second", and possibly issue corrective actions. You can't do that unless you are holding the wheel and are paying attention to what the car is doing! But all that means is that the car is not really driving itself; it is however dealing with the boring, predictable parts of the job and leaving t…

The boring predictable parts of driving (lane keeping and speed keeping) are already basically unconscious muscle memory. It’s watching others for surprises that takes the mental energy.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#252

Post by rayiner » Thu, Mar 21, 2019, 2:04 AM UTC

slg wrote: post #173

rayiner wrote: post #153

Earlier quoted context omitted.

I think you're missing the forest for the trees. We're not even at the point where we're comparing safety rates. Can a Model Y drive through DC? Can it ignore traffic signals in favor of following hand signals from a uniformed traffic officer, or follow the directions of a construction worker having traffic in turns through a road that they've decided to turn into one lane? Can it deal with an unannounced presidentia…

As far as I'm aware, Tesla has never claimed their cars will be able to handle all those situations. "Self driving car" is a vague term that has different definitions depending on who you ask. You are considering "self driving" as level 5 autonomy and Tesla probably considers it somewhere between 3-4 autonomy. I don't think it makes sense to get angry because their definition of the phrase is not the same as your def…

That’s just called “driving” and if the car can’t handle those situations it’s not a “self driving car.” It’s just a driver assistance feature.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#253

Post by saalweachter » Thu, Mar 21, 2019, 2:11 AM UTC

leggomylibro wrote: post #214

skwb wrote: post #172

Earlier quoted context omitted.

You don't unit test a NN. You can unit test certain functions, but fundamentally this is an integration test. This is why serious scientific training is needed to understand these complex systems when health and safety are on the line.

"Autonomous vehicle integration test track" could make for a great setting in a spy thriller. The villain could own the megacorporation which makes the cars, and the heroes could find evidence of their evil plot among the sprawling acres of labs and potemkin streets. But then, in the distance, the sound of revving engines... Seriously though, I wonder if that sort of physical test track will become popular. You would…

> "Autonomous vehicle integration test track"

Just say "the 101", it's shorter.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#254

Post by jaimex2 » Thu, Mar 21, 2019, 2:13 AM UTC

jtaft wrote: post #230

Can they simulate their sensors input pretty well? Do they use car driving simulators (video-game like) to see what would happen in similar scenarios? Of course, real world testing would still be necessary.

When a customer disengages or corrects AP that periods sensor data is recorded and uploaded to Tesla. The uploads are curated and submitted into the neural network training data-set.

They basically have an infinite amount of real world data coming in.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#255

Post by gwbas1c » Thu, Mar 21, 2019, 2:17 AM UTC

falcolas wrote: post #55

craftyguy wrote: post #45

Earlier quoted context omitted.

Sure, if you construe the meaning of 'fixed' to mean 'roll back all events and time', which I don't think anyone expects. They probably fixed the software bug.

I think that it's important to note that "we'll fix bugs via live patches" is a deadly decision when it comes to self driving software. Just discussing the incident in terms of a "software bug" does a disservice to the severity of the issue.

Remember, autopilot and autosteer are not self-driving. Tesla is very explicit that the driver must remain alert, and supervise, at all times.

That being said, drowsy driving is a thing, and it's very easy to fall asleep behind the wheel. The car really needs a better strategy to handle this situation.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#256

Post by gwbas1c » Thu, Mar 21, 2019, 2:19 AM UTC

The video really doesn't convey what happened. It's not really clear when the driver took over.

My Model 3 is very odd when going through forks like this. It just has trouble picking one side of the lane or the other. Yesterday I got a little scared when it swerved back and forth while trying to figure out how to take an exit.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#257

Post by dsfyu404ed » Thu, Mar 21, 2019, 2:19 AM UTC

slg wrote: post #169

rayiner wrote: post #155

Earlier quoted context omitted.

> You are setting up a false dichotomy. You don't have to choose between Autopilot and paying attention. The video above is an obvious example. It didn't result in an accident because the driver was paying attention and intervened. If you have to pay attention, then you might as well be driving.

This is just a silly argument to me. Would anyone suggest that cruise control is worthless because you still have to steer the car?

Well if marketing says you get a CNC mill and what you really get is three power feeds it's not worthless but it's still not what you were promised.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#258

Post by mr_toad » Thu, Mar 21, 2019, 2:21 AM UTC

kirse wrote: post #184

erobbins wrote: post #138

I think Tesla eventually has a catastrophic accident and is sued and/or criminally prosecuted into oblivion. I feel sorry for the people who are going to have to die for this to happen. This trope that humans are bad drivers is, in general, crap. Humans are very good drivers. The US has 7.3 deaths per billion km driven. This means if you drive 50km a day, every day, you are (essentially) guaranteed to die... after 75…

I feel sorry for the people who are going to have to die for this to happen. I try to explain this to friends who are far too optimistic on self-crashing car technology. Self-driving cars (SDC) ultimately trade one class of problems that result in death (human-attention deficit) for another class of growing issues (sensor malfunctions/incapabilities, software defects). Ultimately SDC deaths end up as bugs/features on…

> it would have been 50x cheaper to simply build a fully modernized high-speed rail network

Where do you get that number from?

Re: After 6 months of working fine, Tesla software update drives at barriers again

#259

Post by toomuchtodo » Thu, Mar 21, 2019, 2:25 AM UTC

CamperBob2 wrote: post #116

toomuchtodo wrote: post #94

Earlier quoted context omitted.

You can still save lives with automotive safety systems and not have a dumpster fire of an SDLC process. There is a spectrum between one firmware update a year and "f * it we'll do it live".

I think a big step forward would be addressing something you mentioned earlier, with respect to documenting what changed in a given update. "Bug fixes" doesn't belong in a Spotify changelog, much less one for a product made by Boeing or Tesla. It almost seems like a right-to-repair issue, where manufacturers are going out of their way to avoid documenting how their products actually work for fear of losing control ov…

You make excellent points, and I think the unfortunate answer is regulation will be required.

Re: After 6 months of working fine, Tesla software update drives at barriers again

#260

Post by maxerickson » Thu, Mar 21, 2019, 2:27 AM UTC

rayiner wrote: post #153

slg wrote: post #106

Earlier quoted context omitted.

I don't want to dismiss your whole point, because it is certainly valid, but that isn't really the issue here. It is entirely possible for bugs like this to exist in the self driving tech and for Tesla to be correct in their claims that Autopilot is on average safer than a human driver. It is obviously troubling to see self driving cars run into solid and stationary objects, but human drivers do that all the time too…

I think you're missing the forest for the trees. We're not even at the point where we're comparing safety rates. Can a Model Y drive through DC? Can it ignore traffic signals in favor of following hand signals from a uniformed traffic officer, or follow the directions of a construction worker having traffic in turns through a road that they've decided to turn into one lane? Can it deal with an unannounced presidentia…

From what I understand, it can't even negotiate a traffic light.

Like here's a recent article swooning over it coming out sometime soon:

https://electrek.co/2018/12/09/tesla-autopilot-soon-traffic-...

The article there also says that the more capable version will require new hardware, which isn't something they have been admitting very readily over the years.