Tesla FSD drives down railroad tracks, gets hit by train

[Zoidz]

Except that in your case, it is fatally flawed.

Prove it.

Sponsored

 

[Donald Stanfield]

Except no one here is anti-technology. Keep grasping at straws…

Anti-technology is my favorite accusation from Teslastans on here. You must have a wrinkle-free brain to suggest a forum full of people driving one of the most advanced vehicles in the world is anti-technology.

 

[COdogman]

Anti-technology is my favorite accusation from Teslastans on here. You must have a wrinkle-free brain to suggest a forum full of people driving one of the most advanced vehicles in the world is anti-technology.

Exactly. Because if you’re not ok with self driving tech being rushed to market without any evidence it’s safe other than the statements of a pathological liar, you don’t TRULY love technology

 

[Zoidz]

Anti-technology is my favorite accusation from Teslastans on here. You must have a wrinkle-free brain to suggest a forum full of people driving one of the most advanced vehicles in the world is anti-technology.

Exactly. Because if you’re not ok with self driving tech being rushed to market without any evidence it’s safe other than the statements of a pathological liar, you don’t TRULY love technology

Exactly. I'm not dubious of Tesla FSD for me if/when I was using it, because I'm responsible and used it "supervised". My problem is the unacceptable risk, however slight, that it presents to other motorists and pedestrians because some a-hole decides to fall asleep, play video games, or f*** their significant other while FSD is driving.

And Elon makes light of FSD sex distraction, essentially condoning it, joking "should have seen it coming".

 

[SteveInBend]

My problem is the unacceptable risk, however slight, that it presents to other motorists and pedestrians...

Hey! Let's not forget cyclists, too.

 

[docwhiz]

Exactly. I'm not dubious of Tesla FSD for me if/when I was using it, because I'm responsible and used it "supervised". My problem is the unacceptable risk, however slight, that it presents to other motorists and pedestrians because some a-hole decides to fall asleep, play video games, or f*** their significant other while FSD is driving.

And Elon makes light of FSD sex distraction, essentially condoning it, joking "should have seen it coming".

I've used FSD on and off (mostly off) since supervised became available.
The most common problem (and the reason I don't use it much) is "phantom braking" where the car will suddenly apply the brakes for no reason.
It's impossible to predict and impossible to prevent even when fully attentive. It's clearly dangerous.
There are lots of other dangerous things it does but the braking is the worst and is why I don't trust it.
Every once in a while I'll try it again if I'm feeling lucky but it usually screws up within a few minutes.

 

[docwhiz]

Hey! Let's not forget cyclists, too.

Sometimes it sees them, sometimes not.

 

[COdogman]

Exactly. I'm not dubious of Tesla FSD for me if/when I was using it, because I'm responsible and used it "supervised". My problem is the unacceptable risk, however slight, that it presents to other motorists and pedestrians because some a-hole decides to fall asleep, play video games, or f*** their significant other while FSD is driving.

And Elon makes light of FSD sex distraction, essentially condoning it, joking "should have seen it coming".

It feels like bizarro world. I love technology - especially automotive technology. And some day ADAS systems will be amazing and I look forward to it.

I'm not sure if those who defend Elon's constant lies about FSD/ Autopilot realize they are not HELPING by lying about it's current capabilities and safety. If anything their behavior will backfire and set its adoption back years...

It's also disgusting to me that some openly dance around admitting they are fine with people dying as part of the experiment. Often people who didn't even know they were included...We really live on the worst timeline

 

[Dave Cundiff]

Not trying to start any more fires here, but here's Fred Lambert's take on the claim that FSD is already "26 times safer" than a human driver:

https://electrek.co/2025/06/16/bloomberg-most-embarassing-report-tesla-waymo-self-driving/

***

Tesla's comparison of FSD crash rates with human-driver crash rates appear to be "apples-to-oranges" comparisons for multiple reasons.

One of the more important is that Tesla's data reflect only crashes in which the airbags or seat-belt pretensioners are deployed. That reflects maybe 18% of total crashes, inflating FSD's supposed safety by a factor of more than five.

My understanding is that when FSD realizes a crash is imminent, it returns control to the human driver and that crash is then attributed to the human rather than to FSD. This also inflates the perceived safety of FSD. I don't have a way of quantifying this bias.

Also, open highways (where most of us prefer to deploy more of the Driver Assistance technologies) have a much lower crash rate per mile driven than urban areas. Adjustment for roadway type would be important as part of any comparison between FSD safety and human safety.

An additional point is that all weather conditions and traffic conditions go into the human-driver denominator and numerator, while FSD's denominator only includes the miles where the human driver has decided to engage FSD.

So I'll wait for unbiased experts to evaluate the safety of Tesla's FSD (and anyone else's).

We'll eventually learn more about DOGE's impact on the number and independence of safety experts at NHTSA. Getting good comparisons may take a while.

***

In medical trials, where the goal is to find whether the new thing is better than the old thing (or better than nothing), experts do "intention to treat" analyses and also adjust for any group-to-group differences in the underlying risk for bad outcomes.

FSD has even more at stake (positively and negatively) than most medical trials. It appears to deserve at least as rigorous a statistical approach.

Best to all!

 

[JM.]

Except no one here is anti-technology.

And nobody claimed they were. You should work on reading the actual words, not adding your own assumptions and incorrect interpretations of them.

 

[COdogman]

And nobody claimed they were. You should work on reading the actual words, not adding your own assumptions and incorrect interpretations of them.

That's an interesting suggestion coming from someone who regularly shares "incorrect interpretations" and "assumptions" about the safety of FSD.

I guess irony isn't dead yet

 

[JM.]

That's an interesting suggestion coming from someone who regularly shares "incorrect interpretations" and "assumptions" about the safety of FSD.

 

[pamalabama]

Yup. Shit like this is why responsible manufacturers who are using self driving on non freeway streets are using radar/lidar in addition to vision. Lidar would detect this obstacle ahead and not try to drive through it. Vision didn’t recognize it as a hazard, and this is what you get. Terrifying how much vision only systems are relying on the processing and recognition of images to keep the driver safe instead of actual object detection like Waymo does.

Lidar has problems with fog. Lidar's main advantage is seeing things at night time that cameras do not see.

Thin objects are a problem too because lidar resolution is not high enough

As far as why someone would do this? He turned onto train tracks when the train is using the other track.

We had just a few weeks ago a guy who flipped over his tesla in "self driving" mode and crashed into a tree. He went to the hospital. He even shared the data of the accident which shows him disengaging FSD, steering it into the tree and flipping the car over. He denies doing it and he believed the data showed he was innocent

FSD will not cross a train track for robotaxi launch. Waymo does not cross train tracks

Tesla reports FSD accidents but likely if this one is not much of an accident. I don't know if telemetry will be triggered

 

[pamalabama]

Easiest way to help make sense of whether FSD did this is to ask the guy where he was headed. The only reason FSD would have done this is if it thought the train tracks were a shortcut

 

[JM.]

Easiest way to help make sense of whether FSD did this is to ask the guy where he was headed. The only reason FSD would have done this is if it thought the train tracks were a shortcut

A good point (and your other post was full of great points too that will trigger the anti-ADAS people). Waymos are known for taking SAFE but unusual shortcuts. The interesting part of that is that they were not told to do this, and somehow "learned" to do it. One example is that one drove through the Waymo storage/charging/service facility to get to the other side. It shouldn't do that, but was also perfectly safe. I'd pay extra for that tour!

Sponsored