NHTSA report shows Tesla Autopilot led the pack in crashes, but the data has gaps

On the face of it, data from the first year of the National Highway Traffic Safety Administration’s project to track the safety of advanced driver assistance systems look terrible for Tesla. Its electric vehicles were involved in 70% of the reported crashes involving Level 2 technologies, 60% of those resulting in serious injuries and nearly 85% of the fatal ones.

The data released early Wednesday were collected under the federal regulator’s Standing General Order issued last June, requiring automakers to report the most serious crashes that involved Level 2 ADAS, which need a human driver to remain fully engaged in the driving task. NHTSA is also tracking crashes involving fully automated vehicles — none of which are currently available to consumers.

There are five levels of automation under standards created by SAE International. Level 2 means two functions such as adaptive cruise and lane keeping are automated and still have a human driver in the loop at all times. Level 2 is an advanced driver assistance system, and has become increasingly common in new vehicles. 

Tesla topped the ADAS list for all the wrong reasons: 273 reported crashes, three with serious injuries, and five deaths. Honda trailed far behind Tesla with 90 crashes and one fatality, while most other manufacturers reported just a handful. Nissan reported none at all.

So does that mean Tesla owners should trade in their Model 3 with Autopilot for a Nissan Leaf, and its own Level 2 ADAS, called ProPilot?

It’s a more complicated question than one might think. The way the Order is worded, the technologies that Tesla has deployed, and the sheer number of Tesla vehicles on the road mean that its vehicles may not be quite as dangerous as the numbers suggest.

For a start, there are more ADAS-equipped Teslas on the road (about 830,000) than vehicles from other manufacturers, although Nissan isn’t far off, with 560,000.

Tesla’s Autopilot can also be used on a variety of roads, unlike Nissan’s ProPilot and GM’s SuperCruise systems, which are limited to highways. Without knowing the number of miles driven with each ADAS system in operation, and where, it is impossible to compare their relative levels of safety — or how each might contrast with accident rates under full human control.

The Order required manufacturers to report all the incidents they knew about, but most vehicles on the road do not have remote telematics that send vehicle data back to the factory. Manufacturers of these cars were reliant on consumer complaints (which comprised the majority of reports), law enforcement contacts or media stories, all of which may not have accurately reported whether their ADAS systems were in use.

Tesla, on the other hand, knows exactly which vehicles were using Autopilot when they crashed, as its vehicles have cellular and Wi-Fi connections that automatically report vehicle data when a crash occurs. Almost all its crash reports were sourced from such telematics, compared to just nine from Subaru, four from GM, three from Lucid and one from Honda.

Finally, the Order required makers to include data on crashes that they were made aware of beginning 10 days after the Order was served last June. In Tesla’s case, that apparently included crashes stretching back to 2019, including three of its five fatal crashes, and all three of its serious ones. (It’s unclear why Tesla was only notified of those accidents months or years after they happened.) Apart from Tesla, only Honda reported a couple of crashes from before June 2021.

While all these variables seem to point in the same direction — a relative over-reporting of Tesla crash data, and under-reporting of crashes involving other car-makers — their impact is impossible to quantify from the NHTSA data alone. Perhaps all Level 2 systems are more dangerous than human drivers alone, due to driver inattention. Or it could be the case that Tesla’s Autopilot as deployed is in fact less competent and more dangerous than rival ADAS technologies.

“The data released today is a good start, but it doesn’t provide an apples-to-apples comparison of advanced vehicle safety,” said National Transportation Safety Board Chair Jennifer Homendy. “What NHTSA provided was a ‘fruit bowl’ of data with a lot of caveats, making it difficult for the public and experts alike to understand what is being reported. Independent analysis of the data is key to identifying any safety gaps and potential remedies.”

The final word on Autopilot will have to wait for NHTSA’s separate, ongoing and recently expanded investigation into Autopilot, which could potentially lead to a recall. In the meantime, drivers with Level 2 systems in their cars would be well advised to heed NHTSA’s advice — “no commercially available motor vehicles today are capable of driving themselves.”