NHTSA Tesla Report (2024) [pdf] | Hacker News Reader

NHTSA Tesla Report (2024) [pdf]

static.nhtsa.gov

37 pointsKingOfCoders36 comments

Penny Arcade had this comic, a few days ago (about the CyberTruck): https://www.penny-arcade.com/comic/2024/04/22/cyberyuck

Are the pedals really being pop-riveted back into place? Tesla isn't known for build quality but that is just horrible.

what might be an acceptable solution here? just curious

Bolt + locktite. Then refinished/repaint/treat the underlying metal to prevent future corrosion from new hole. The pedal is now in metal-to-metal contact with the pan, in a damp environment. That means differential metal issues.

I don't really see the problem with pop rivets in that application. What's the issue?

Pop rivets loosen, loosen in a way that cannot be tightened. They should use engineered bolts/nuts secured with locktite. If those do slip, they can re-tightened. Pop rivets must be cut off and replaced. You won't see pop rivets on any other cars, not since the 1940s. ("Blind" rivet are different than pop rivets)

If they are pop-riveting, then I assume they are doing this on the cheap and are also not treating/protecting the metal. They are probably just drilling through without worry about future corrosion issues.

What do you think the difference between a pop rivet and a blind rivet is? There are different types of blind rivet fastening systems with different properties, but the words are absolutely used interchangeably. Also, blind rivets are used extensively in automotive applications.

I do think this fix looks quite shoddy in general, but I don't think a bolt/nut would be a better solution in that location.

How is this related to the linked NHTSA report, which so far as I can tell is about issues with Tesla's Autopliot system and not the accelerator pedal issue? Also, I get that comics aren't the place to go for thoughtful analysis and hard hitting investigative reporting, but I'm struggling to extract anything meaningful from that comic aside from an inane "DAE Tesla drivers bad?".

Even the associated article seems to be less about the Cybertruck (it only gives a brief synopsis of the accelerator pedal issue) and more about Bethesda.

>I'm struggling to extract anything meaningful from that comic aside from an inane "DAE Tesla drivers bad?".

That is indeed the entirety of it.

Cybertruck is this generation's Hummer H1, and those jokes went on for over a decade.

> extract anything meaningful from that comic

Speaking only for myself, the only thing I usually am interested in "extracting" from comics, is humor.

If it is not funny to you, then that makes me sad, but I think a number of other folks seemed to find some utility in it.

The biggest reveal here is that they provide one of the clearest confirmations that Tesla has been lying about their crash statistics. Apparently they only log crashes with airbags and generally from police reports we know this is only 18% of crashes. They found and confirmed several instances of crashes that don’t show up in Teslas logs at all.

It does not sound accurate to me that only 18% of crashes appear in police reports?

I think he means 18% of crashes that the police report have an airbag deployment.

Yes thanks that is what I meant.

I don't know the percentage, but the majority of minor collisions never result in a police report. The parties involved just deal with their insurers. When a guy scratched my bumper at a stoplight I didn't waste time calling the police.

Wouldn’t that be absent from other car manufacturer’s data as well though?

Many minor crashes aren't in police reports at all. Either it involves only one party, or the police just straight up won't come out.

The number one search result for "Autopilot" on HN is still this 2017 article claiming a 40% reduction in crash rate: https://news.ycombinator.com/item?id=13437905 The subsequent revelations of the many flaws in this analysis never had one tenth the fanfare of course.

40% reduction in crash rate, not fatalities. It seems dubious that they could get fatality statistics wrong and get away with it. Creative statistics feels plausible with “crashes”.

Ah you're right, thanks. Edited my OP.

"Another crash occurred in March 2023 in North Carolina and involved a 2022 Model Y. The vehicle was traveling at highway speed when it struck a minor pedestrian exiting a school bus. The pedestrian was evacuated by air to a hospital for treatment of serious injuries. Based on publicly available information, both the bus and the pedestrian would have been visible to an attentive driver and allowed the driver to avoid or minimize the severity of this crash."

These systems need to be banned from public roads until a comprehensive test suite is developed which verifies their ability to drive safely in all conditions.

If autopilot should be banned because preventable tragedies occur when autopilot is used, doesn't it follow that automobiles should be banned because preventable tragedies occur when automobiles are used?

Since cell phones themselves directly contribute to automobile-related preventable tragedies as well, shouldn't we ban cell phones too?

After all, there are zero metrics about any product or service that matter other than whether or not they cause preventable deaths.

Let's get rid of knives, the internet, all drugs (since at least some drugs cause preventable deaths), the video games, and the satanic music too if we're already going on a maternalistic, emotional banning spree of anything that has ever caused preventable deaths.

Would there be any benefit to adding self-drive sensors and recording to a vehicle but not allow it control, then just record everything the car senses while being driven by the public. You'd then have a massive amount of data to feed into an expert system or similarly-constrained ML system. Take that data over a couple of years, across the world's drivers, and you'd have a lot of useful stuff.

The carmakers probably have something like this for training FSD but would having a lot more data help build a better self-drive?

Every OEM dreams about this, MobilEye built and deployed it already (REM [0]).

[0] https://www.mobileye.com/technology/rem/

[dupe]

More discussion: https://news.ycombinator.com/item?id=40170135

https://news.ycombinator.com/item?id=40171540

Was the fatal FSD Beta crash the mention in the news?

"Frontal Plane Struck Vehicle / Object / Person in travel path: 143"

It’s disappointing to see the US continuing to foster negligent driving attitudes by attributing blame to a brand name (Autopilot) instead of attributing blame where it belongs: inattentive drivers.

Regardless of what assistive driver tool is used, it doesn’t change the fact that the driver is the only human-thinking operator behind the wheel of a 2-4 ton vehicle. Sensors and cameras can fail. Software can crash. Humans are the only realistic failsafe right now for consumer driver assist.

Attempting to shift blame to a driver assist continues our decades-long path of refusing to admit that we have a careless and underskilled driver epidemic.

The amount of people driving with a cellphone and a thumb actively moving is absurd

The problem is this tool (Autopilot) advertises itself as something other than what it is. (Driver assist.) The report also states that this tool in particular behaves differently from its peers in that it discourages the driver from participating in corrective driving while it's engaged, leading to less attentive use of the tool.

See my other reply.

A name change will not fix inattentive drivers. You’re falling into exactly the same trap I’ve described: excusing everything but the driver.

Let's say you are right that there is an epidemic of careless and underskilled drivers. Is it then ok to give those people a drive assist system named "autopilot" and "full self driving" with strong wink and nudge that people can drive with less attention ? Regardless of legal responsibility, with your assumption, it seems it's totally fair to criticize Tesla for making an already terrible problem worse.

You’re ignoring almost all of what I typed.

The fact of the matter is, the name does not matter. Multiple warning messages get displayed that get ignored by drivers. You’re doing exactly what I called out: blaming everything but the driver.

Are you proposing a mere name change is all that is required to improve reading comprehension on warning labels?

People are generally proposing all of at least (1) a name change, (2) a stop to the false claims and marketing and (3) better driver monitoring. Some people are proposing additional things, like (4) changing the way it reacts to corrections in order to not discourage active involvement, or (5) limiting the situations in which it can be enabled, or even (6) changing how it is turned on.

If there are issues with getting drivers to pay attention to what they're doing, would you say that's likely to be helped or hindered by a tool which allows them to ignore the road entirely 99% of the time?

Sure, I think ADAS simply existing is responsible for encouraging inattentive driving. I don’t have a solution for this, though, since almost every automaker has some form of ADAS available now.

Many of us are developers or in the line of work. We should know first-hand how difficult it is to guide end-user behavior.

Reply on news.ycombinator.com