Although the driver’s seat was found vacant and the driver was found in the left rear seat, the available evidence suggests that the driver was seated in the driver’s seat at the time of the crash and moved into the rear seat postcrash. Specifically, residential security video showed both the driver and passenger getting into the front seats prior to driving away from the residence. In addition, the EDR data showed active accelerator pedal inputs consistent with driver activity in the 5 seconds prior to the impact with the tree, and that the driver’s seat belt was connected at the time of the crash. Finally, the steering wheel examination conducted by the NTSB Materials Laboratory indicated an impact to the upper left quadrant, consistent with the driver loading the steering wheel during a frontal crash.
The entire journey appears to have lasted less than a minute (9:07pm is mentioned both as the time the car left the driveway, and as the time of the crash), so it's unlikely the driver jumped into the back seat as a stunt while driving.
https://www.ntsb.gov/investigations/AccidentReports/Reports/...
The NTSB might have been trying to use neutral language for uncertain events, like how "the position changed" could mean intent or just a force of nature. I don't know what the right interpretation is.
It's surprising how fast I got out of the car via back seats without even thinking.
(only thought was - door on my side is potentially blocked, get out...)
(And yeah, I'm on record saying that it's extremely unlikely autopilot commanded the lane change in that Bay Bridge accident. It just doesn't work like that, the blinkers came on simultaneous with the motion. That's a human driver for sure.)
And notably, Musk did not chime in to blame the driver like he always does when the driver is at fault, which is very strong circumstantial evidence that FSD was engaged.
Look, the clearly most likely scenario here is that the driver got confused, overrode the AP, and then failed to maintain awareness as they commanded the sudden lane change. Obviously that's not what people want to hear, but it's the best explanation. I've pulled AP out and changed lanes manually, probably rapidly, on a bunch of occasions. All it takes is to do that when not paying attention...
All your logic was equally applicable to the accident in the linked article, and it was all wrong there too. (Also... Musk didn't "blame the driver" in the linked case? Not sure what you're citing. Tesla generally doesn't say anything about their customers' accidents, for some very obvious reasons.)
Just wait. All I'm telling you is that I drive with this product every day and I'm not seeing FSD behavior in that video. If it turns out I'm wrong then I'll sheepishly admit it. But I know which direction the spin runs and my BS detector is pretty good.
And the fact that this was media disinformation won't matter much, because the damage to the Tesla brand and the idea of self driving cars in the public's collective consciousness has already been done, and won't easily be undone by these newly revealed facts.
"A lie makes it halfway around the world before the truth even gets its boots on."
https://www.khou.com/article/news/local/tesla-spring-crash-f...
And even in that article, there's an addendum a couple of months later stating that a person was seen getting in the driver's seat and that the NTSB reenactment wasn't able to engage the autosteering when trying to replicate the crash.
I don't know what more you want from the press. They reported what they were told by the authorities. And then when new information came in, they reported that as well.
Possibly it's the media's job to question when law enforcement comments seem implausible?
Self driving cars were billed as the cool new thing. Then Tesla started charging people for "Full Self Driving" with a system that regresses from basic LKA that was shipping in the mid 2000s in terms of safety by using public roads for beta testing an intentionally hamstrung* L2-billed-as-L5 driving stack.
It's frustrating to watch as someone working in the AV space, and it's silly to act like this one case was isolated enough that the fact it was wrong changes any of that reality. We know Tesla's stack has killed people: it became normalized once the first few times it happened they were able to get away with victim blaming. By 2021 it was already accepted, this incident was already business as usual.
*humans just need eyes so self driving cars just need cameras
Cite? Everyone likes to throw around the word "safety" as a qualitative thing, but the only data we have points to the system being extremely safe.
The "damage to the Tesla brand and the idea of self driving cars" wasn't caused solely by this one incident.
> "A lie makes it halfway around the world before the truth even gets its boots on."
It's not a lie to say cigarettes kill people by lung cancer, even if it's later proven that one smoker who died in some study actually got lung cancer from a different cause.
This is one of those things that always struck me as a weird, poor argument... how is autonomous driving supposed to suddenly make everyone want to stop owning their cars and use taxis instead?
The insurance companies that underwrite them, on the other hand.
Also, empirically, traditional car companies are actually investing heavily in self-driving tech. Their capabilities actually surpass Tesla’s.
Other causes of human driver failure include driving while tired or exhausted, falling asleep at the wheel, driving while angry or upset, getting bored while driving, using the car's infotainment system while driving, using a mobile phone while driving, getting distracted by passengers, smoking pot while driving, being a recklessly immature teenager (or a grown-up idiot), lacking the bare minimum of driving skills that every driver on the road is supposed to have, and so on. The list of causes of human driver failure is long. There are a lot of horrifically dangerous human drivers -- look around you next time you're on the road.
Notably, machines are immune to all these human failure modes.
I'm looking forward to the day in which cars drive themselves well enough to rid the roads of so much dangerous human driving.
Meanwhile, excessive speed and intoxication account for more than 50% of all fatalities. You can solve this problem without having to wait for an AI car, it's probably morally justified to do even if you think those AI cars will arrive "any minute now."
Finally.. I am not looking forward to the day when a parent puts their kids in an AI car and does not accompany them to their destination. People will invent new failure modes for you.
Amazingly it’s worse than that. He was also on pills that would react with the alcohol to make things worse.
I’ll say I’m very surprised by the areal photo in the report. I was under the impression the driver (man or machine) had gone a real distance at speed before the accident. Which did lead some credence in my mind a driver assist system had been helping it along if the driver was slow to react/inattentive.
He could have pulled out of that driveway, hit the gas, and it happened. The trip had barely begun. You could do this crash easy in a 1960s sedan, no ADAS needed.
A lot of our driving in unusual conditions require our experience and reasoning about the world.
If you build a train that does what I need without requiring copious amounts of my spare time, I might take it. If you don't then your only "motivation" boils down to "reducing options until the train is all that is left."
People need infrastructure that meets their demands. If it works, they'll be plenty motivated.
The report states:
The frontal impact with the tree resulted in a power loss of the car’s 12-volt system, which runs the non-traction power systems. During normal operation, the front door latches operate electronically with the pull of the interior lever. In the event of a 12-volt system power loss, the interior front doors open as usual using the interior door handles. The rear doors also have both electronic and mechanical latches; however, mechanically opening the rear door during a power loss requires additional steps. According to the owner’s manual, during a loss of 12-volt system power, a rear-seated occupant must locate a small cutout in the carpet beneath the seat cushions and pull the mechanical release cable tab toward the center of the vehicle to manually open the rear door. Inspection of the door latches and locking hardware was limited by postcrash fire damage.
Edit: Must add that this assumes the front door was jammed and couldn't be opened mechanically.
Someone totaled my Model 3 on my birthday a few years ago and my front door was smashed. I smelled smoke and thought I had a battery fire (turns out it was smoke from the airbags). I crawled to the back seat and tried to open my door but it wouldn't open due to power loss. I ended up breaking the window because in my state of mind, I couldn't remember that the manual door override was available. People always say there is a manual backup, but even if it works, it's really hard to remember that during a terrible accident where time is an issue. Luckily I carry a knife in my car with one of those window breaking points.
I can't imagine trying to do that drunk.
Do you mean it would have been lucky in a different situation?
Because the NTSB cannot establish that conclusively given the almost total lack of physical evidence remaining after the vehicle fire.
It does not mean that a vehicle defect does not actually exist from the NTSB's point-of-view. It just means that the NTSB lacks the physical evidence to conclude anything there.
(I personally find the necessary "additional steps" required for exiting the vehicle from the rear seats in the event of a power loss troubling.)
More broadly, this NTSB report is being misinterpreted by many here.
The "probable causes" (not "causes") and "lessons learned" (tellingly distinct from the NTSB's more traditional "safety recommendations") established in the final report are exactly tailored to the physical evidence and physical facts that did remain independently of Tesla (the company) - and no more.
The actual title of the HN submission is, in my view, inappropriately written by excluding the word "probable".
This reads like a terrible design.
If you have a Tesla with these stupid rear door locks, make sure your children know how to open them.
Tesla's vehicles are not capable of self-driving and, at all times, the human driver is driving the vehicle as both Autopilot and FSD Beta are partial automated driving systems that require a human driver fallback at all times.
The attentiveness required of the human driver with a partial automated driving system is equivalent (on a systems-level) to if the vehicle was not equipped with any automated driving system at all.
My hot take is that it shouldn't be possible to accelerate that quickly unless you actually apply the pedal to 100%.
Also, I think the accelerator and brake pedals should have a different feel. Maybe they should be made of different materials so there is clear tactile feedback.
If the driver was indeed wearing a seatbelt, how does that person end up in the left back seat in a frontal crash? Now I have questions about that seat belt.
Especially considering this comment in the report: In addition, the EDR data showed active accelerator pedal inputs consistent with driver activity in the 5 seconds prior to the impact with the tree, and that the driver’s seat belt was connected at the time of the crash.
So there was a driver pushing the accelerator, wearing a seatbelt (assumption) and ending up in the rear seat behind the drivers' seat? Did the driver simply plug in the seatbelt without putting it on? Or did the driver survive the crash, get into the rear seat, and was unable to get out in time?
>Recently, the NTSB recommended that NHTSA require all new vehicles to be equipped with passive vehicle-integrated alcohol impairment detection systems, advanced driver monitoring systems, or a combination thereof, which are capable of preventing or limiting vehicle operation if driver impairment by alcohol is detected.
>A vehicle technology-based solution, such as intelligent speed adaptation (ISA), can reduce speeding. The NTSB has recommended that NHTSA incentivize passenger vehicle manufacturers and consumers to adopt ISA systems by, for example, including ISA in the New Car Assessment Program.
The full list is here https://www.ntsb.gov/investigations/Pages/Investigations.asp... You can see it's been a while since since they investigated a Tesla.
Two people killed in fiery Tesla crash with no one driving https://news.ycombinator.com/item?id=26852399 736 comments
NTSB: Tesla Model S in Crash Couldn't Have Been Using Autopilot - https://news.ycombinator.com/item?id=27112010 199 comments
Cops “almost 99.9% sure” Tesla had no one at the wheel before deadly crash https://news.ycombinator.com/item?id=26871001 53 comments
No one was in driver’s seat in fatal Tesla crash https://news.ycombinator.com/item?id=26866754 85 comments
Texas police to demand Tesla crash data as Musk denies autopilot use https://news.ycombinator.com/item?id=26869962 18 comments
