Tesla Safety Report: Response to Waymo's Data Request

Tesla Releases Detailed Data on Full Self-Driving Safety

In response to calls for greater transparency, Tesla has recently unveiled comprehensive data regarding the performance and safety characteristics of its advanced driver-assistance system. This publication follows remarks made by Waymo’s co-CEO, Tekedra Mawakana, at TechCrunch Disrupt, urging companies to share more detailed fleet information.

Miles Driven Between Collisions

According to a newly established section on its website, Tesla reports that vehicles utilizing its Full Self-Driving (Supervised) software in North America are averaging approximately 5 million miles driven before experiencing a significant collision. Furthermore, the data indicates around 1.5 million miles are driven before a less severe collision occurs.

These figures represent a considerably improved safety record when contrasted with national averages as reported by the National Highway Traffic Safety Administration (NHTSA). Tesla’s interpretation of NHTSA data suggests that the average driver encounters a major collision every 699,000 miles and a minor collision every 229,000 miles.

Previous Safety Reports and Robotaxi Trials

Tesla has been issuing “vehicle safety reports” on a quarterly schedule. However, these reports have consistently faced criticism for lacking sufficient detail. Limited information has also been made public concerning the safety performance of Tesla’s Robotaxi trial, currently underway in Austin, Texas, which still necessitates the presence of a safety driver.

Waymo’s Safety Claims

Waymo, currently a leading robotaxi provider in the U.S. based on deployment and customer reach, has published data demonstrating its vehicles are approximately five times safer than those driven by humans. With regard to pedestrian safety, Waymo claims a twelvefold improvement. At last month’s Disrupt conference, Mawakana was questioned about other companies contributing to road safety.

“I am currently unaware of other companies providing this level of data regarding their fleet performance,” Mawakana stated, pointedly avoiding mention of Tesla.

The Importance of Transparency

Mawakana further emphasized the responsibility of companies deploying autonomous vehicles. “If a driver is removed from direct control, and oversight is provided remotely with the ability to intervene, transparency regarding fleet operations is paramount. A lack of transparency, in my opinion, indicates a failure to adequately prioritize road safety,” she explained.

Waymo has not yet issued a comment regarding whether Tesla’s newly released data addresses Mawakana’s concerns.

Focus Shift to Full Self-Driving (Supervised)

Past criticisms of Tesla’s quarterly safety reports centered on their emphasis on Autopilot, a less sophisticated driver-assistance system than the Full Self-Driving (Supervised) software – often referred to as FSD. It’s important to note that despite its name, FSD does not currently achieve full autonomy.

Autopilot is primarily designed for highway use, where crash rates are generally lower, particularly when considering minor incidents.

New Data Breakdown for FSD

Tesla has now provided a comprehensive breakdown of data related to FSD. The company asserts that drivers utilizing FSD experience major collisions approximately every 2.9 million miles, compared to the national average of 505,000 miles per major collision, according to NHTSA data. For minor collisions, FSD users average 986,000 miles between incidents, while the national average is around 178,000 miles.

Defining Collision Severity

For the first time, Tesla is also clarifying its definitions of “major” and “minor” collisions.

The car manufacturer is aligning its definitions with the Federal Motor Vehicle Safety Standards, specifically 49 C.F.R. § 563.5. “Major collisions” are defined as those involving the deployment of a vehicle’s airbags or other pyrotechnic safety devices.

Furthermore, Tesla states that any collision occurring within five seconds of FSD being actively engaged will be included in the dataset, even if the driver disengages the system or it aborts prior to impact.

Data Updates and Future Reporting

“This methodology ensures that our reported collision rates for FSD (Supervised) encompass not only collisions occurring during active system control, but also scenarios involving driver disengagement or system-initiated aborts immediately before an impact,” Tesla clarifies.

In its frequently asked questions section, Tesla commits to updating this data quarterly, utilizing a rolling twelve-month aggregation of miles and collisions to reflect current trends. The company indicates it will not publicly release data on injury rates, as this information is automatically collected from vehicles.

“Tesla prioritizes objective, programmatic metrics such as collision frequency and airbag deployment rates. Airbag deployments provide a reliable indicator of collision severity,” the company concludes.