New video of Tesla crash demonstrates the problem of semi-automated driving • Hacker News Board

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#241

Post by modeless » Thu, Jan 12, 2023, 5:13 AM UTC

paulryanrogers wrote: post #232

modeless wrote: post #197

Earlier quoted context omitted.

Automatic switching between FSD and Autopilot is not any kind of UX problem. There is no reason the driver needs to be aware of it. The driver does not need to take any action at all. The only difference is the type of road. But which software was in use is relevant here because people are trying to blame this on FSD. Whenever you're on a freeway, Autopilot is used automatically instead of FSD, so FSD had nothing to…

When responsible for overseeing a multi-ton vehicle, I like to know how it will behave. Cruise control and lane assist I understand. I need to know which one is in control and how it will behave if they can respond differently.

There is no difference in behavior when switching from FSD to Autopilot or vice versa. And there is never any ambiguity about which is active because it is completely determined by road type which you're obviously aware of while driving.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#242

Post by Gustomaximus » Thu, Jan 12, 2023, 5:13 AM UTC

axelfontaine wrote: post #205

throwaway892238 wrote: post #200

I'm still waiting for autonomous-only lanes. For extremely long and boring highways (usually full of trucks), this just seems like the logical conclusion of all self-driving tech to me. Dedicate one lane to self-driving, have the vehicles all on autonomous mode, and have them keep enough distance so their braking performance prevents them from crashing in an unexpected stop. You should be able to literally just sleep…

We have this technology today. It's called trains.

It doesn't as trains don't have the convenience of your own car where the stops are.

If you had a train you could drive your car onto it would be more equivalent.

Also train lines have far less coverage than highways.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#243

Post by justapassenger » Thu, Jan 12, 2023, 5:17 AM UTC

somethoughts wrote: post #65

I think what's interesting here is it's likely the first instance where Tesla FSD has been involved in an accident which affected other drivers. [Edit] From the video the Tesla is making a lane change and stopping simultaneously which means there could be a case of the Tesla FSD/driver doing a unsafe lane change.[1] Most of the time FSD just wrecks the Tesla itself or injures the driver of the Tesla (i.e. running int…

It’s very unlikely it’s a first time FSD caused accident for others. People using FSD allow it to do reckless things, just to see “if it’ll figure it out”. You can cause accident, without being involved in it, especially if you drive in unpredictable way.

Like this one - if driver would stop sudden breaking and moved forward, cars behind him could have still crashed.

That’s the thing about testing on the public roads - there are many ways you can affect other users.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#244

Post by int_19h » Thu, Jan 12, 2023, 5:17 AM UTC

paulryanrogers wrote: post #228

int_19h wrote: post #168

Earlier quoted context omitted.

They should, and I said as much. It does not absolve the other person from responsibility for not following the required safety protocol. The car in front of them might have had to brake for a perfectly legitimate reason. I mean, suppose that the other driver was driving while drunk. They might have arrived to their destination safely without the asshole in front, but that's not an excuse against a DUI ticket.

How do you maintain safe distance from those in adjacent lanes? Especially if they may come from behind you, suddenly change lanes in front, then break hard?

You do not, but that wasn't the hypothetical posited by OP.

In the actual case with Tesla, the driver of the car that was in front of the pile-up was not at fault IMO. But there were a bunch more cars behind, and at least some of those slammed into each other because they didn't maintain proper distance, according to the police.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#245

Post by sfmike » Thu, Jan 12, 2023, 5:17 AM UTC

throwaway892238 wrote: post #200

I'm still waiting for autonomous-only lanes. For extremely long and boring highways (usually full of trucks), this just seems like the logical conclusion of all self-driving tech to me. Dedicate one lane to self-driving, have the vehicles all on autonomous mode, and have them keep enough distance so their braking performance prevents them from crashing in an unexpected stop. You should be able to literally just sleep…

Makes sense. We have HOV lanes which reward for being good for the environment. Even better is an autonomous lane that rewards not adding latency to the prisoners dilemna, seems the ultimate altruistic move, someone wanting to not impede or slow others.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#246

Post by sfmike » Thu, Jan 12, 2023, 5:18 AM UTC

soraminazuki wrote: post #225

axelfontaine wrote: post #205

Earlier quoted context omitted.

We have this technology today. It's called trains.

Reminds me of how I used to build railways everywhere instead of roads in Sim City. But surely that wouldn't work out in real life?

In Tokyo at least in the proper city it's easy to feel you're fine with just JR + subways.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#247

Post by osigurdson » Thu, Jan 12, 2023, 5:18 AM UTC

I think the insistence on having data for every possible situation for training purposes is indicative of the problem. Humans only require a small amount of training and can extrapolate this to many situations.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#248

Post by a_bonobo » Thu, Jan 12, 2023, 5:18 AM UTC

Eisenstein wrote: post #217

axelfontaine wrote: post #205

Earlier quoted context omitted.

We have this technology today. It's called trains.

All of Elon Musk's grand ideas for revolutionizing terrestrial transportation (that aren't laughably unworkable like miles long vacuum enclosures) are just 'trains but much smaller and with tires so that we lose all efficiencies and gain all of the drawbacks'. So this fits perfectly.

The difference to trains in Musk (and Silicon Valley's overall view) is that with automated cars, you don't have to share your personal space with other people (the poors!). It's easy to make fun of the overall project but there's elitism behind it - have your private space at the cost of the public.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#249

Post by onethought » Thu, Jan 12, 2023, 5:19 AM UTC

threeseed wrote: post #33

kiratp wrote: post #25

This looks like the vehicle tried and failed to get the driver to engage and pulled over best it could. There is no shoulder there so… Let’s wait to get more info as to what the driver was doing and if they were incapacitated. The FSD beta is pretty aggressive on making sure the driver is paying attention via both steering sensors, the in-cabin camera and the touch screen.

> pulled over best it could If that is its best then FSD has no business being on the road. It was erratic and didn't give clear intent to other cars what it was trying to do.

Because this is the first car pile up we've ever seen.

Interesting because a computer was involved we are incensed. "No business being on the road"... but the millions of other car crashes "oh yeah, well humans make mistakes!"

Tesla's claims of being "safer than human drivers" is as much a testament to their technology as it is to how poor human drivers are.

Re: New video of Tesla crash demonstrates the problem of semi-automated driving

#250

Post by modeless » Thu, Jan 12, 2023, 5:19 AM UTC

somethoughts wrote: post #239

modeless wrote: post #202

Earlier quoted context omitted.

Wrong. You can plainly see in the video (and the hundreds of others) that as soon as the car enters the freeway it switches to Autopilot automatically. The easiest way to tell is the red lines in the FSD visualization disappear when it switches to Autopilot. The transition is completely smooth and the driver doesn't have to know or care at all, so it's not surprising that they were mistaken. But it's relevant here be…

My take is that you are probably right in that it is pretty easy for Tesla to place all the blame on the driver of the Tesla and weasel out of any legally required support for the driver via TOS fine legal print in the short term. But after the 10th (or maybe 100th) scenario where Tesla FSD is at fault but Tesla scapegoats responsibility onto the driver who is branded by Tesla's legal team as a dumb/irresponsible/clu…

> you are probably right in that it is pretty easy for Tesla to place all the blame on the driver of the Tesla

Whoa now, I didn't say that at all! I'm just talking about FSD vs. Autopilot here. It's entirely possible that Autopilot is at fault! I am curious to see what the investigation will uncover. My point is simply that this situation has nothing to do with FSD. It could have happened four years ago as easily as yesterday, and anyone trying to tie this to recent developments in FSD is being misleading.