TL;DR: Self-Driving Teslas Rear-End Motorcyclists, Killing at Least 5
Brevity is the spirit of wit, and I am just not that witty. This is a long article, here is the gist of it:
- The NHTSA’s self-driving crash data reveals that Tesla’s self-driving technology is, by far, the most dangerous for motorcyclists, with five fatal crashes that we know of.
- This issue is unique to Tesla. Other self-driving manufacturers have logged zero motorcycle fatalities with the NHTSA in the same time frame.
- The crashes are overwhelmingly Teslas rear-ending motorcyclists.
Read our full analysis as we go case-by-case and connect the heavily redacted government data to news reports and police documents.
Oh, and read our thoughts about what this means for the robotaxi launch that is slated for Austin in less than 60 days.
In Boca Raton, I’ve seen no evidence that the self-driving tech was inactive. According to the government, it is reported as a self-driving accident, and according to the driver in his court filings, it was active.
Insanely, you can slam on the gas in Tesla’s self-driving mode, accelerate to 100MPH in a 45MPH zone, and strike another vehicle, all without the vehicle’s “traffic aware” automation effectively applying a brake.
That’s not sensationalist. That really is just insanely designed.
FTFA:
If the guy smashes the gas, just like in cruise control I would not expect the vehicle to stop itself.
The guy admitted to being intoxicted and held the gas down… what’s the self driving contribution to that?
I know what’s in the article, boss. I wrote it. No need to tell me FTFA.
TACC stands for Traffic Aware Cruise Control. If I have a self-driving technology like TACC active, and the car’s sensor suite detects traffic immediately in front of me, I would expect it to reduce speed (as is its advertised function). I would expect that to override gas pedal input, because the gas pedal sets your maximum speed in cruise control, but the software should still function as advertised and not operate at the maximum speed.
I would not expect it to fail to detect the motorcyclist and plow into them at speed. I think we can all agree that is a bad outcome for a self-driving system.
Here’s the manual, if you’re curious. It doesn’t work in bright sunlight, fog, excessively curvy roads (???), situations with oncoming headlights (!?!), or if your cameras are dirty or covered with a sticker. They also helpfully specify that “The list above does not represent an exhaustive list of situations that may interfere with proper operation of Traffic-Aware Cruise Control,” so it’s all that shit, and anything else - if you die or kill somebody, you have just found another situation that may interfere with proper function of the TACC system.
https://www.tesla.com/ownersmanual/2012_2020_models/en_us/GUID-50331432-B914-400D-B93D-556EAD66FD0B.html#%3A~%3Atext=Traffic-Aware+Cruise+Control+determines%2Cmaintains+a+set+driving+speed.
So do you expect self driving tech to override human action? or do you expect human action to override self driving tech?
I expect the human to override the system, not the other way around. Nobody claims to have a system that requires no human input, aside from limited and experimental implementations that are not road legal nationwide. I kind of expect human input to override the robot given the fear of robots making mistakes despite the humans behind them getting into them drunk and holding down the throttle until they turn motorcyclists into red mist. But that’s my assumption.
With the boca one specifically, the guy got in his car inebriated. That was the first mistake that caused the problem that should never have happened. If the car was truly self driving automated and had no user input, this wouldn’t have happened. It wouldn’t have gone nearly 2.5x the speed limit. It would have braked long in advance before hitting someone in the road.
I have a ninja 650. We all know the danger comes from things we cannot control, such as others. I’d trust an actually automated car over a human driver always, even with limited modern tech. The second the user gets an input though? zero trust.
The driver being drunk doesn’t mean the self-driving feature should not detect motorcycles. The human is a fallback to the tech. The tech had to fail for this fatal crash to occur.
If the system is advertised as overrriding the human speed inputs ( traffic aware cruise control, it is supposed to brake when it detects traffic, regardless of pedal inputs), then it should function as advertised.
Incidentally, I agree, I broadly trust automated cars to act more predictably than human drivers. In the case of specifically Teslas and specifically motorcycles, it looks like something is going wrong. That’s what the data says, anyhow. If the government were functioning how it should, the tech would be disabled during the investigation, which is ongoing.