All while Waymo is expanding to more and more cities, including Detroit where it will deal with snow and ice. Waymo is years ahead of Tesla in the self-driving race. It's possible Tesla never succeeds in launching a truly self-driving car.
It feels extremely disingenuous to have a video of burned Teslas (from vandalism) at the top of the article... why do this? Is there any explanation for this except malice? It has absolutely nothing to do with the safety of robotaxis.
Isn't it spicy enough to just report on the safety issues from Robotaxis?
Also:
> But several of the Austin crashes occurred while the vehicles were moving slowly or stationary, one incident involved contact with a fixed object in a parking area. Analysts say this suggests the system's perception and decision-making may not be giving monitors enough time to react, a key issue NHTSA has previously flagged in other FSD-related investigations.
Interesting for sure. This is also what some of the FSD influencers see when they test the limits (especially with parking with small obstacles).
The Electrek article (https://electrek.co/2025/10/29/tesla-robotaxis-keep-crashing...) contains more information. It's 4 crashes in around 250,000 miles. But Tesla redacts most information in these reports unlike e.g. Waymo, so the information is limited. If Tesla wants people to trust them regarding safety, this is not the way to go.
Personally, I'd be interested in how often the safety drivers had to intervene. But I assume we'll never get that information.
The root of Tesla's problems here is Musk's decision to drop LIDAR.
The reasoning, I think, was that humans can drive using sight and a little bit of sound, so an AI should be able to do this too.
Humans can do this because we have a very rich well-developed world model that allows us to fill in the gaps. We don't do it perfectly but we can do it decently well.
Modern AIs, or at least the ones small enough to be run on smaller machines that are economical to put in cars, don't have a rich world model like that. They're doing stimulus response backed by a database. That's going to break down at all kinds of edge cases.
The way to compensate for this is to give the car superhuman senses like LIDAR. The car is much dumber than a person but it can perceive its environment orders of magnitude better than a person, which compensates well enough that it has a chance of driving at least as well as a person.
As a massive advocate of FSD (and someone who's currently running 14.1.4, their very latest FSD build) it is absolutely in no way ready for unsupervised FSD. It still makes silly mistakes, and the latest build is terrified of leaves and will swerve across lane dividers to desperately avoid a leaf blowing into it's path.
I love FSD, I use it for 99% of my driving, and when it's working right it's an incredibly technology that overall makes my driving safer, but there's absolutely areas of weakness that every FSD user knows it cannot be trusted in under any circumstances and you must closely supervise and be ready to take over at any time.
Elon already stopped talking about the robotaxi and is now onto the Optimus bot, which should buy him at least another year or two of swindling investors.
Perhaps Musk is so busy making up shit about what is happening in the UK and trying to foment a civil war there, that he doesn't have time to manage Tesla?
We don’t know the details. It could be that the human drivers were in control at the time of the incidents and caused some of them. It could also be that other cars driven by humans caused them.
Web sites hosting these clickbait articles have zero incentive to make things sound less dramatic.
An obvious tell is that they’ll use the word “crash” for a Tesla bumping a parking bollard.
Didn't this happen before with a safety monitor on their phone? I seem to remember another Robo taxi company and it hit someone who was crossing the street.
14 comments
[ 2.4 ms ] story [ 26.0 ms ] threadThis is in the article right below a picture of the safety monitor in the passenger seat…
Isn't it spicy enough to just report on the safety issues from Robotaxis?
Also:
> But several of the Austin crashes occurred while the vehicles were moving slowly or stationary, one incident involved contact with a fixed object in a parking area. Analysts say this suggests the system's perception and decision-making may not be giving monitors enough time to react, a key issue NHTSA has previously flagged in other FSD-related investigations.
Interesting for sure. This is also what some of the FSD influencers see when they test the limits (especially with parking with small obstacles).
Personally, I'd be interested in how often the safety drivers had to intervene. But I assume we'll never get that information.
The reasoning, I think, was that humans can drive using sight and a little bit of sound, so an AI should be able to do this too.
Humans can do this because we have a very rich well-developed world model that allows us to fill in the gaps. We don't do it perfectly but we can do it decently well.
Modern AIs, or at least the ones small enough to be run on smaller machines that are economical to put in cars, don't have a rich world model like that. They're doing stimulus response backed by a database. That's going to break down at all kinds of edge cases.
The way to compensate for this is to give the car superhuman senses like LIDAR. The car is much dumber than a person but it can perceive its environment orders of magnitude better than a person, which compensates well enough that it has a chance of driving at least as well as a person.
I love FSD, I use it for 99% of my driving, and when it's working right it's an incredibly technology that overall makes my driving safer, but there's absolutely areas of weakness that every FSD user knows it cannot be trusted in under any circumstances and you must closely supervise and be ready to take over at any time.
Ah yes, the old “crashed while stationary bug”. Hard to fix, that one.
Web sites hosting these clickbait articles have zero incentive to make things sound less dramatic.
An obvious tell is that they’ll use the word “crash” for a Tesla bumping a parking bollard.