Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

latimes.com

Post reply on HN
Print view View original

81–90 of 163 posts

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#81

Post by lttlrck »

mcot2 wrote: post #60
kadoban wrote: post #53

Earlier quoted context omitted.

If it's not true, Tesla should be overjoyed to share raw data (not processed through their marketing department) then.

I don’t think they even really have a traditional marketing department. The beta release notes are probably something you should check out. Latest: Improved object detection network architecture for non-VRUs (e.g. cars, trucks, buses). 7% higher recall. 16% lower depth error, and 21% lower velocity error for crossing vehicles. New visibility network with 18.5% less mean relative error. New general static object netwo…

Percentages are quite useless without the absolute numbers as a reference...

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#82

Post by mcot2 »

Pyramus wrote: post #59
mcot2 wrote: post #38

Earlier quoted context omitted.

Because its not true at all. Most major Tesla incidents of any kind make headlines from fires to accidents on AP. You should do some basic research on the topic. https://www.theverge.com/platform/amp/2021/10/21/22738834/te... The FUD is _insane_ with Tesla.

So why doesn't Tesla share raw data with the public, or regulators? I cannot think of better PR for Tesla, and autonomous driving in general.

They do and it is easily googled.

https://www.tesla.com/VehicleSafetyReport

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#83

Post by sroussey »

Jxl180 wrote: post #70
mcot2 wrote: post #57

Actual user of the beta here. Happy to answer any questions that are honest. My opinion is let the technology develop in the open, warts and all.

My delivery isn’t until June but I’ve held off on buying FSD (maybe I’ll enable it for the month when I do a long trip), but I’m worried about this fantom braking issue I’ve heard about.

I would get an older one. Newer ones lack the expensive radar component. Camera-only can be disabled by kids with a Q-beam (bright flashlight).

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#84

Post by wpietri »

mcot2 wrote: post #57

Actual user of the beta here. Happy to answer any questions that are honest. My opinion is let the technology develop in the open, warts and all.

Help me understand how you think your opinion entitles you to gamble with other people's lives.

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#85

Post by mcot2 »

jondwillis wrote: post #48
staunch wrote: post #22

I'm a big fan of Tesla but I'm fairly certain that, at some point, they're going to voluntarily refund everyone the money they paid for self-driving. And, at some later point, they will probably make self-driving work but on a future generation of HW/SW. Genuine question: if this happens, will it kill Tesla to refund all of this money? I don't pay close enough attention to their financials to know myself. I assume (a…

Software isn’t really a problem. What hardware changes do you think Tesla will need to make to deliver on FSD, whether that ends up being actual level 5 or not? it is also not that expensive to retrofit most new hardware, especially if it is just new silicon.

They have already done a hardware upgrade to my car from HW2.5->3, the nivida based solution to their own custom silicon. The $10,000 (now, $5000 when I bought it) includes all hardware upgrades.

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#86

Post by wongarsu »

jimkleiber wrote: post #54
thinkmassive wrote: post #23

Note that there are two “FSD Beta” programs: 1. The paid software unlock for “Full Self Driving” (currently in “beta”) available to everyone, which provides automatic driving from on-ramp to off-ramp, and a handful of parking features like auto-park and summon: https://www.tesla.com/support/autopilot#usingautopilot 2. The invite-only “FSD Beta” that unlocks the above FSD features on city streets, which is available t…

I feel confused why 1) they would call both programs by the same name (arguing that an acronym is pretty much the same name) and 2) why the government even allows them to call it Full Self Driving, when there are levels of autonomous driving and I imagine most would assume Full means the highest level, which, Tesla has not yet achieved from everything I've read.

> I imagine most would assume Full means the highest level

USB 2.0 got away with Full Speed/High Speed (with high being more than full)

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#87

Post by TigeriusKirk »

mcot2 wrote: post #71
TigeriusKirk wrote: post #62

Earlier quoted context omitted.

I believe Tesla testing FSD on open roads is probably going to cost lives. I think they're aware of the game they're playing and they're taking more risks than is responsible. I also believe the driver in that video doesn't have the level-headedness to be testing anything anywhere. Which is a criticism both of the driver and of Tesla for letting people like that run the software on the open road.

I think it is a calculated risk to get to autonomy faster and thus save lives.

Yes, I do think that's what their game is. Which I'm not inherently opposed to. But I think that at present they're taking more risk than I'm comfortable with having on our public streets.

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#88

Post by jazzyjackson »

mcot2 wrote: post #67
ModernMech wrote: post #34

I really thought the breaks would have been slammed on Tesla's shipping of autonomous car software after the incident where the guy was decapitated by a tractor trailer.

At that time the system was merely lane keep with adaptive cruise control. Any system would have had a similar result at that time.

should have had emergency auto-brake. As I understand it the radar either didn’t see the trailer or filtered it out as stationary - camera was unable to distinguish white steel from white sky - should have been a wake up call that current tech doesn’t have the sensory bandwidth to know if something is blocking the road.

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#89

Post by wpietri »

intrasight wrote: post #58
mygoodaccount wrote: post #20

Slightly off topic, but it's fun to see company culture reflected in the algorithm behaviour. There are thousands of clips of Tesla's Fatal Self Driving being far too aggressive for American roads ( https://nitter.net/taylorogan/status/1469404579439824899 ). Meanwhile, companies like AutoX have the problem of being far too conservative for Asian roads ( https://www.youtube.com/watch?app=desktop&v=TFEvkmvIjVo ). AutoX…

This I've long thought the be the Achilles' heel of autonomous. Even skilled human drivers have a hard time calibrating their style to the local situation. I remember the first time I drove as a passenger through Manhattan with my father driving, and couldn't get my head around his change in driving style. Was a side of him I'd never seen before - driving aggressively like a cab driver. I'm sure it came from his year…

For sure. Driving on a test track is a technical problem. Driving in traffic is a social problem. As you noted, how one drives in traffic is not just about getting somewhere, but about communicating and negotiating with other drivers around you.

It's one of the things that makes me think that true self-driving may be in the AGI bucket.

Re: Alarmed by Tesla’s self-driving test, state legislators demand answers from DMV

#90

Post by kadoban »

mcot2 wrote: post #60
kadoban wrote: post #53

Earlier quoted context omitted.

If it's not true, Tesla should be overjoyed to share raw data (not processed through their marketing department) then.

I don’t think they even really have a traditional marketing department. The beta release notes are probably something you should check out. Latest: Improved object detection network architecture for non-VRUs (e.g. cars, trucks, buses). 7% higher recall. 16% lower depth error, and 21% lower velocity error for crossing vehicles. New visibility network with 18.5% less mean relative error. New general static object netwo…

They _absolutely_ have a marketing department, PR whatever you want to call it.

Those numbers don't look useful for judging safety. How would one compare those against anything else?

Post reply on HN