Tesla reports another Robotaxi crash(electrek.co) |
Tesla reports another Robotaxi crash(electrek.co) |
Yet somehow they claim that old versions, using old hardware, on arbitrary roads, using untrained customers as safety drivers somehow average 2.9 million miles per collision in non-highway environments [2], a ~72.5x difference in collision frequency, and 5.1 million miles per collision in all environments, a ~175x(!) difference in collision frequency, when their reporting and data are not scrutinized.
I guess their most advanced software and hardware and professional safety drivers just make it 175x more dangerous.
[1] https://techcrunch.com/2025/05/20/musk-says-teslas-self-driv...
[2] https://www.tesla.com/fsd/safety
[3] https://www.forbes.com/sites/alanohnsman/2025/08/20/elon-mus...
[3.a] Tesla own attorneys have argued that statements by Tesla executives are such nonsense that no reasonable person would believe them.
I'm not sure what the guys in the taxis with their hands on the arm rest do. I guess they have a button that either stops the car or connects it to a remote control operator?
People often don't report minor accidents. Someone scrapes a pole without causing enough damage to hit their insurance deductible, are they going to file a police report? Mostly not. And then the older number had that in it and the newer one doesn't.
But the number for human drivers works like the old number. They're dividing miles driven by reported accidents. On top of that, they're using the average -- by miles -- which isn't the same as the median, and in particular that will over-represent drivers who drive the most miles, who are disproportionately professional drivers.
Ultimately, Tesla has two problems going on here:
1. Their crash rate is 2x that of Waymo.
2. They redact a lot of key information, which complicates safety assessments of their fleet.
The redactions actually hurt Tesla, because the nature of each road incident really matters: EVERY traffic incident must be reported, regardless of fault (even if it's a speeding car from the other direction that hits another vehicle which then hits the robotaxi - yes, that's actually in one of the Waymo NHTSA incident reports). When Tesla redacts the way they've been doing, it makes it very difficult to do studies like https://www.tandfonline.com/doi/full/10.1080/15389588.2025.2... which show how much safer Waymo vehicles are compared to humans WHEN IT COMES TO ACTUAL DAMAGE DONE.
We can't get that quality of info from Tesla due to their redaction practices. All we can reliably glean is that Tesla vehicles are involved in 2x the incidents per mile compared to Waymo. https://ilovetesla.com/teslas-robotaxi-dilemma-navigating-cr...
(The one thing I would like to see done differently here is including an error interval.)
Of the Tesla accidents, five of them involved either collisions with fixed objects, animals, or a non-injured cyclist. Extremely minor versions of these with human drivers often go unreported.
Unfortunately, without the details, this comparison will end up being a comparison between two rates with very different measurement approaches.
Austin has relatively low miles so the confidence intervals are wider but not too far from what they show for other cities
Please use the original title, unless it is misleading or linkbait; don't editorialize.
> and other self driving technologies
I mean, this isn't self-driving. It has a safety driver.
https://www.yahoo.com/news/articles/two-self-driving-waymos-...
We need far higher quality data than this to reach meaningful conclusions. Implying conclusions based upon this extrapolation is irresponsible.
In the past it took a lot less to get the situation fixed… and these were horrendous situations! [1][2] And yet tesla is a factor of 10 worse!
[1] https://en.wikipedia.org/wiki/Ford_Pinto
[2] https://en.wikipedia.org/wiki/Firestone_and_Ford_tire_contro...
This one is misleading both in that 8 "crashes" is statistically insignificant to draw conclusions as to its safety compared to humans, but also because these 'crashes' are not actually crashes and instead a variety of things, including hitting a wild animal of unknown size or potentially minor contact with other objects of unspecified impact strength.
They make other unsubstantiated and likely just wrong claims:
> The most critical detail that gets lost in the noise is that these crashes are happening with a human safety supervisor in the driver’s seat (for highway trips) or passenger seat, with a finger on a kill switch.
The robotaxi supervisors are overwhelmingly only the passenger seat - I've never actually seen any video footage of them in the driver seat, and Electrek assuredly has zero evidence of how many of the reported incidents involved someone in the driver seat. Additionally, these supervisors in the passenger seat are not instructed to prevent every single incident (they arent going to emergency brake for a squirrel) and to characterize them as "babysitting to prevent accidents" is just wrong.
This article is full of other glaring problems and lies and mistruths but it's genuinely not worth the effort to write 5 pages on it.
If you want some insight on why Fed Lambert might be doing this, look no further than the bottom of the page: Fred gives (sells?) "investment tips" which, you guessed it, are perpetually trying to convince people to sell and short Telsa: https://x.com/FredLambert/status/1831731982868369419
Feel free to look at his other posts: it's 95% trying to convince people that Telsa is going bankrupt tomorrow, and trying to slam Elon as much as possible - sometimes for good reasons (transphobia) but sometimes in ways that really harms his credibility, if he actually had any
Lambert has also been accused of astrotrufing in lawsuits, and had to go through a settlement that required him to retract all the libel he had spread: https://www.thedrive.com/tech/21838/the-truth-behind-electre...
That same source also touches on Fred and Seth's long history of swinging either side of the bandwagon in attempts to maximize personal gain off bullshit reporting. And basically being a massive joke in automotive reporting.
The owner of Eletrek, Seth Weintraub, also notably does the same thing: https://x.com/llsethj/status/1217198837212884993
With 7 reported crashes at the time, Tesla’s Robotaxi was crashing roughly
once every 40,000 miles [...]. For comparison, the average human driver
in the US crashes about once every 500,000 miles.
This means Tesla’s “autonomous” vehicle, which is supposed to be the future of safety,
is crashing 10x more often than a human driver.
A responsible journalist with half a clue would mention that, and tell us how that distorts the numbers. If we correct for this distortion, it’s clear that the truth would come out in Tesla’s favor here.
Instead the writer embraces the distortion, trying to make Tesla look bad, and one is left to wonder if they are intentionally pushing a biased narrative.
Using your own personal experience, it should be obvious that trivial fender benders are more common than once per lifetime but significantly less common than one every couple of years.
Does anyone know what the cite for this might be? I'm coming up empty. To my knowledge, no one (except maybe insurance companies) tallies numbers for fender bender style accidents. This seems like a weirdly high number to me, it's very rare to find any vehicle that reaches 100k miles without at least one bump or two requiring repair.
My suspicion is that this is a count of accidents involving emergency vehicle or law enforcement involvement? In which case it's a pretty terrible apples/oranges comparison.
Waymo has a huge head start, and it is evident that the "fully autonomous" robotaxi date is far behind what Elon is saying publicly. They will do it, but it is not as close as the hype suggests.
The data is just not there for us outsiders to make any kind of case, and thats the skimping out crucial baseline we need.
Every time when he has the choice to do something conservative or bold, he goes for the latter, and so long as he has a bit of luck, that is very much a winning strategy. To most people, I guess the stress of always betting everything on red would be unbearable. I mean, the guy got a $300m cash payout in 1999! Hands up who would keep working 100 hour weeks for 26 years after that.
I'm not saying it is either bad or good. He clearly did well out of it for himself financially. But I guess the whole cameras/lidar thing is similar. Because it's big, bold, from the outset unlikely to work, and it's a massive "fake it till you make it" thing.
But if he can crack it, again I guess he hits the jackpot. Never mind cars, they are expensive enough that Lidar cost is a rounding error. But if he can then stick 3d vision into any old cheap cameras, surely that is worth a lot. In fact wasn't this part of Tesla's great vision - to diversify away from cars and into robots etc. I'm sure the military would order thousands and millions of cheapo cameras that work 90% as well as a fancy Lidar - while being fully solid state etc.
That he is using his clients as lab rats for it is yet another reason why I'm not buying one. But to me this is totally in character for Musk.
The only real test will be who creates the best product, and while waymo seems to have the lead it's arguably too soon to tell.
It didn't work out though and now multi sensor systems are eating their lunch.
The radar they had really couldn't detect stationary objects. It relied on the doppler effect to look for moving objects. That would work most of the time, but sometimes there would be a stationary object in the road, and then the computer vision system would have to make a decision, and unfortunately in unusual situations like a firetruck parked at an angle to block off a crash site, the Tesla would plow into the firetruck.
Given that the radar couldn't really ever be reliable enough to create a self driving vehicle, after he hired Karpathy, Elon became convinced that the only way to meet the promise was to just ignore the radar and get the computer vision up to enough reliability to do FSD. By Tesla's own admission now, the hardware on those 2016+ vehicles is not adequate to do the job.
All of that is to say that IMO Elon's primary reason for his opinions about Lidar are simply because those older cars didn't have one, and he had promised to deliver FSD on that hardware, and therefore it couldn't be necessary, or he'd go broke paying out lawsuits. We will see what happens with the lawsuits.
Price is a factor. I’ve been using the free self driving promo month on my model Y (hardware version 4), and it’s pretty nice 99% of the time.
I wouldn’t pay for it, but I can see a person with more limited faculties, perhaps due to age, finding it worthwhile. And it is available now in a $40k vehicle.
It’s not full self driving, and Waymo is obviously technologically better, but I don’t think anyone is beating Tesla’s utility to price ratio right now.
Simplifying things doesn't always make things easier.
[1] https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/...
[2] https://www.nhtsa.gov/crash-data-systems/crash-report-sampli...
[1] Source: I've hit a garbage can and told no one. Until this moment.
It goes seem like a high number to me - in 30 years of pretty heavy driving I've probably done about 500k miles and I've definitely had more than one incident. But not THAT many more than one, and I've put 100k miles on a few vehicles with zero incidents. Most of my incidents were when I was a newer driver who drove fairly recklessly.
1/500k miles that includes the interstate will be very different from the rate for an urban environment.
Something like this: https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEh...
I don't see why you say that.
> My point is that the data naturally has error margins that are clearly large enough to make drawing concrete conclusions impossible.
I don't understand this one either.
He "painted himself into a corner" is the more accurate expression when one is the cause of their own problem
Nevertheless, getting big money donations out of politics would be a big improvement.
Despite that, the article and the public (the target of the hit piece that encourages people to endanger themselves with a system that has not been demonstrated to be safe with the direct intent of enriching the owners of Tesla) directly refute Tesla's ridiculous claims demonstrating they are off by multiple orders of magnitude using basic mandatory data reporting for their Robotaxi program which is using systems more advanced, fine-tuned, geofenced, with professional safety drivers (thus we can only reasonably assume that their normal system is worse), but which actually has scrutinized reporting requirements.
And yet now you argue that the entity fabricating ridiculous claims for their own enrichment, Tesla, is not only not responsible, but target of the hit piece, the ones that clearly and debunked Tesla's claims as deceptive, are not only responsible for refuting it but are responsible for demonstrating a level of rigor that is unimpeachable when the original fabricated claim lacks even the elements of rigor we expect out of your average middle school science fair, let alone a literal trillion dollar company.
Talk about double standards.
Also I don't think that's correct; that's a ton of driving! I strongly suspect the number you're citing is the number of miles an average American spends in a road vehicle, not actually driving it. But that counts the same "car-mile" multiple times for all the occupants, when the statistic we're arguing about right now is about the vehicle, not the occupants.
> For months, Tesla’s robotaxis in Austin and San Francisco have included safety monitors with access to a kill switch in case of emergency — a fallback that Waymo currently doesn’t need for its commercial robotaxi service. The safety monitor sits in the passenger seat in Austin and in the driver seat in San Francisco
They've been seen doing this at crime scenes and in the middle of police traffic stops. That speaks volumes too.
> presented with a set of options and they choose one
> they send a physical human to drive the car.
Those all sound like "controls" to me.
"Fleet response can influence the Waymo Driver's path, whether indirectly through indicating lane closures, explicitly requesting the AV use a particular lane, or, in the most complex scenarios, explicitly proposing a path for the vehicle to consider. "
https://waymo.com/blog/2024/05/fleet-response/
So they built new controls that typical vehicles don't have. Then they use them. I fail to see how any of this is "incorrect." It is, in fact, _built in_ to the system from the ground up.
Semantic games aside, it is obviously more incorrect to call them "completely self driving" especially when they "ask for help." Do human drivers do this while driving?
How could the richest man in the world give so little back.
Actually, we know that vision alone doesn't work.
Sun glare. Fog. Whiteouts. Intense downpours. All of them cause humans to get into accidents, and electronic cameras aren't even as good as human eyes due to dynamic range limitations.
Dead reckoning with GPS and maps are a huge advantage that autonomous cars have over humans. No matter what the conditions are, autonomous cars know where the car is and where the road is. No sliding off the road because you missed a turn.
Being able to control and sense the electric motors at each wheel is a big advantage over "driving feel" from the steering wheel and your inbuilt sense of acceleration.
Radar/lidar is just all upside above and beyond what humans can do.
The way I see it, he converted his cars' greenness into other people's fumes. So not a net win after all.
https://www.npr.org/2025/12/06/nx-s1-5635614/waymo-school-bu...
Engineering problems aren't limited to a single solution anyhow. Anyone ruling out a camera based solution has very little imagination.
I will also add in my personal experience, while some filters work best together (like imu/gnss), we usually either used lidar or camera, not both. Part of the reason was combining them started requiring a lot more overhead and cross-sensor experts, and it took away from the actual problems we were trying to solve. While I suppose one could argue this is a cost issue (just hire more engineers!) I do think there's value in simplifying your tech stack whenever possible. The fewer independent parts you have the faster you can move and the more people can become an expert on one thing
Again Waymo's lead suggests this logic might be wrong but I think there is a solid engineering defense for moving towards just computer vision. Cameras are by far the best sensor, and there are tangible benefits other than just cost.
> solid engineering defense for moving towards just computer vision
COUNTERPOINT: WAYMO
> Again Waymo's lead suggests this logic might be wrong but I think there is a solid engineering defense for moving towards just computer vision. Cameras are by far the best sensor, and there are tangible benefits other than just cost.
It's not that we ran into problems, it's that the tech didn't deliver what we hoped when we could have used the time to build something better.
But also, if you didn’t get the right result, I don’t care how quickly you didn’t get it.
It's analogous to communications latency. High latencies are very annoying to humans, but below a threshold they stop mattering.
Effective altruism and other New Age garbage pseudo philosophy can't hold a candle to that.
And he's retired, so the money is no longer useful to him for maintaining control over the company he runs or expanding it, which is when it traditionally starts going to charities.
To reveal my own bias / worldview, I loathe and detest Bill Gates in nearly every way and have done so for over three decades. I think he has had a massively negative impact on humanity, mainly by making the computer industry so shitty for 4+ decades but in other more controversial areas as well.
With Elon Musk, while perceiving a number of big faults in the man, I also acknowledge that he has helped advance some very beneficial technologies (like electric vehicles and battery storage). So I have a mixed opinion on him, while with Gates, he is almost all evil and has had a massive negative impact on the planet.
I guess what I'm saying is that when people only fixate on one oligarch, which one they mostly focus on can be quite telling.
If that's your definition of solved, be my guest, but it's a pretty silly one.
Detroit, Denver, Philadelphia and DC this year.
Second of all, they’re using cameras to detect things like signs and trash cans! I don’t know where this misconception came from that Waymo only uses lidar, but it demonstrates a lack of familiarity with the problem space and the solutions
this allows them to tag things that their cars are not recognizing, but then are fed back into the training set and validated. it allows them to "live patch" troublesome sections of road but it's not intended for long term use.
they also use this to close sections of road, such as for car accidents, flooding, etc.
