Tesla has published the most detailed look at the performance and relative safety of its advanced driver-assistance software, just a few weeks after Waymo’s co-CEO Tekedra Mawakana at TechCrunch Disrupt called on companies to release more data.
On a new section of its website, Tesla claims that in North America, owners using the company’s Full Self-Driving (Supervised) software are driving around 5 million miles before a major collision and around 1.5 million miles before a minor collision.
That’s a far lower rate than the national average based on statistics provided by the National Highway Traffic Safety Administration (NHTSA). That data shows people get in a major collision every 699,000 miles, and a minor one every 229,000, at least according to Tesla’s interpretation.
Tesla has been releasing “vehicle safety reports” on a quarter-by-quarter basis for a while. But those reports have been repeatedly panned for being insufficient. And Tesla has released almost no information about the safety performance of the Robotaxi trial it’s been running in Austin, Texas, this year, which still has employees in the driver’s seat monitoring for safety reasons.
Waymo, the leading robotaxi company in the U.S. at the moment based on cars deployed and customers served, has published detailed data showing its vehicles are around 5x safer than human drivers, and 12x safer with respect to pedestrians. At last month’s Disrupt conference, Mawakana was asked to name other companies she felt were making roads safer.
“I don’t know who’s on that list, because they’re not telling us what’s happening with their fleets,” said Mawakana, without naming Tesla.
“I think there is a responsibility, if you’re going to put vehicles on the road, and you’re going to remove the driver from behind the wheel, and you’re going to have someone in some other room observing the fleet who can take over their vehicles, it is incumbent upon you to be transparent about what’s happening,” she added. “And if you are not being transparent, then it is my view that you are not doing what is necessary in order to actually earn the right to make the road safer.”
Waymo didn’t immediately respond to a request for comment Friday about whether Mawakana believes Tesla’s new data is sufficient.
One of the repeated criticisms of Tesla’s quarterly safety reports is that it focused on Autopilot, a far less advanced driver-assistance system than the Full Self Driving (Supervised) software, or FSD — which, despite its name, does not make a car fully autonomous. Autopilot was designed to be used on highways, which typically see a lower rate of crashes (when including minor collisions).
Tesla has finally broken out all this data. The new section of Tesla’s website claims that drivers using FSD travel about 2.9 million miles between major collisions, while NHTSA data shows all drivers travel about 505,000 miles per major collision. Tesla claims FSD users drive about 986,000 miles between minor collisions, while NHTSA data shows all drivers travel around 178,000 miles per minor collision.
Tesla is also finally showing how it defines these terms for the first time.
The carmaker is using the Federal Motor Vehicle Safety Standards, specifically 49 C.F.R. § 563.5. Tesla defines “major collisions” as crashes with higher-severity impacts where a vehicle’s airbags “or other non-reversible pyrotechnic restraints” are deployed. The company also says that if FSD was active “at any point within five seconds leading up to a collision event,” then it includes that crash in this dataset.
“This calculation ensures that our reported collision rates for FSD (Supervised) capture not only collisions that occur while the system is actively controlling the vehicle, but also scenarios where a driver may disengage the system or where the system aborts on its own shortly before impact,” Tesla says.
In its FAQ section, Tesla states that it will update the data every quarter and that it will “reflect a rolling twelve-month aggregation of miles and collisions in an effort to remain relevant to recent trends and progress.” The company says it won’t release other information, like injury rates, because it is collecting this data automatically from the vehicles.
“Instead, Tesla focuses on objective and programmatic metrics such as collision frequency and airbag deployment rates. Airbag deployments serve as a reliable proxy for collision severity,” the company writes.
