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

Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

cnbc.com

Post reply on HN
Print view View original

151–160 of 864 posts

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#151

Post by justapassenger »

apnew wrote: post #64

> The FSD Beta system may cause crashes by allowing the affected vehicles to: “act unsafe around intersections, such as traveling straight through an intersection while in a turn-only lane, entering a stop sign-controlled intersection without coming to a complete stop, or proceeding into an intersection during a steady yellow traffic signal without due caution,” according to the notice on the website of the National…

> Does anyone have insights on what QA looks like at Tesla for FSD work? Because all of these seem table-stakes before even thinking about releasing the BETA FSD.

Tesla is not exactly in love with QA. Especially for FSD.

FSD is mainly 2 things: 1. (By far most important) shareholder value creating promise, that's been solved for 6 years according to their CEO. 2. Software engineering research project

What FSD is not is a safety critical systems (which it should be). They focus on cool ML stuff and getting features, with any disregard for how to design, build and test safety critical systems. Validation and QA is basically non-existent.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#152

Post by pxx »

enslavedrobot wrote: post #40
thepasswordis wrote: post #21

It seems like there is a huge disconnect between people who know about FSD from using it and people who know about FSD from things they read on the internet. I have FSD, I use it every single day. I love it. If every car on the road had this the road would be a substantially safer place.

I have the same experience. I think if there was more transparency in the data around FSD related accidents the conversation would be different. The last evidence we have is Tesla stating that there have been no fatalities in 60 million miles of driving under FSD. Pretty good so far.

Regardless of whether or not that's true, that's also not the impressive number you think it is. Human-driven cars have a fatality rate of about 1 per 100 million miles travelled, and that's including drunk drivers, inattentive drivers, and all environments, not just the ones people are comfortable turning on these features in.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#153

Post by bdcravens »

qup wrote: post #117
rootusrootus wrote: post #66

Earlier quoted context omitted.

I have it (subscription, just wanted to try it out). I won't use it again. It confidently attempted to drive me into a rather solid object with very little warning. If every other car on the road had this, I would sell all my cars and forbid family members from getting anywhere near the road. It's nice on the freeway though.

To me it's weird you can voice an opinion like "I won't use it" and then say it's nice on the highway, as if you've given it enough of a trial there to endorse it when you think otherwise it's going to be killing people.

Pretty sure almost crashing into a solid object seems good enough for giving it a failing score.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#155

Post by drowsspa »

rootusrootus wrote: post #50

Meanwhile, the only actual L3 capable car is a Mercedes. Plenty of people will argue that AP is actually superior, but I don't think there's any way to say that for sure, given the tight constraints Mercedes has put on their driver assistance in order to lower the liability enough to make L3 not a bankruptcy event. For certain, however, they are more confident in their technology than Tesla is, otherwise Tesla would…

That Tesla was allowed to test in production without any legal liability with real human lives mostly because of Musk's personal influence is a disgrace

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#156

Post by dawnerd »

yumraj wrote: post #143
apnew wrote: post #64

> The FSD Beta system may cause crashes by allowing the affected vehicles to: “act unsafe around intersections, such as traveling straight through an intersection while in a turn-only lane, entering a stop sign-controlled intersection without coming to a complete stop, or proceeding into an intersection during a steady yellow traffic signal without due caution,” according to the notice on the website of the National…

> Does anyone have insights on what QA looks like at Tesla for FSD work? Yes. An army of Tesla owners perform the QA, in production.

Well first it goes to influencers that say it's perfect and good for stable release no matter what the car does!

But in all seriousness they do have some small team that validates then it goes to employees.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#157

Post by adwn »

saboot wrote: post #78
pirate787 wrote: post #59

Earlier quoted context omitted.

[flagged]

I'm really struck at how this comment has no consideration of the quality or accuracy of current FSD technology. Also, public transit would do the same.

> Also, public transit would do the same.

Does public transit pick me up at my front door and drop me off at any location I choose?

Does public transit take of the minute I leave the house?

Is public transit available 24/7, even in remote areas?

Does public transit wait for me to load 2 kids, a dog, 4 suit cases, and various bags, blankets, and toys?

Does public transit turn around after 2 minutes, bring me back, and wait for 30 seconds because I forgot my wallet on the kitchen counter?

Will public transit still work during a public transit strike? (yes, this actually happens over here in Europe)

No? Well, then you can hardly claim that "public transit would do the same" as a self-driving car.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#158

Post by justapassenger »

sockaddr wrote: post #136
naillo wrote: post #118

Earlier quoted context omitted.

Makes me think karpathys "bet" on vision being the only sensor was maybe misguided

Yup. You need an AGI behind that vision for that premise to work. Even a fruit-fly class AGI would do it.

Flies hit windows and cars. Even much more complex animals cannot manage effectively moving in a complex environment without hitting each other (like flocks of sheep).

There's no data that would support that anything than human level AGI is required to drive cars with how current infrastructure looks like.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#159

Post by Xylakant »

sergiotapia wrote: post #86

I thought emperically, statistically, the number of crashes from FSD was drastically lower than human drivers. Was this number hogwash or is this just lawboys saying the number must be 0 before it's allowed on the road? If so, that's pretty crazy and people will die because of this decision.

The number was problematic for various reasons:

The usual comparison is vs. all cars on the road, but Teslas are comparatively new cars with a lot of recent and expensive safety features not available in older cars. They’re also likely to be maintained better. On top, they’re also a driven by a different demographic which skews accident statistics.

Teslas autopilot can only be enabled in comparatively safe and simple circumstances, yet the comparison is made against all driver in all situations. When autopilot detects a situation it can’t handle, it turns off and hands over to the human who gets a few seconds warning. Human drivers can’t just punt the issue and then crash.

Tesla FSD may be safer than Human drivers for the limited set of environments where you can use it, but last time I checked, the numbers that Tesla published are useless to demonstrate that.

Re: Tesla recalls 360k vehicles, says full self-driving beta may cause crashes

#160

Post by TaylorAlexander »

apnew wrote: post #64

> The FSD Beta system may cause crashes by allowing the affected vehicles to: “act unsafe around intersections, such as traveling straight through an intersection while in a turn-only lane, entering a stop sign-controlled intersection without coming to a complete stop, or proceeding into an intersection during a steady yellow traffic signal without due caution,” according to the notice on the website of the National…

Andrej Karpathy was the AI lead for most of the project and he has talked about the general system design.

They have a set of regression tests they run on new code updates either by feeding in real world data and ensuring the code outputs the expected result, or running the code in simulation.

It does seem worrying that they would miss things like this.

Here’s a talk from Karpathy explaining the system in 2021:

https://youtu.be/aNVbp0WKYzY

Though I don’t recall if he explains the regression testing in this talk, there’s a few good ones on YouTube.

Post reply on HN