Tesla’s closely-guarded driving data has been decrypted for the first time, according to a Dutch government-run forensic lab. The Netherlands Forensic Institute (NFI) said it discovered a wealth of information about Tesla’s Autopilot, along with data around speed, accelerator pedal positions, steering wheel angle and more. The findings will allow the government to “request more targeted data” to help determine the cause of accidents, the investigators said.
The researchers already knew that Tesla vehicles encrypt and store accident related data, but not which data and how much. As such, they reverse-engineered the system and succeeded in “obtaining data from the models S, Y, X and 3,” which they described in a paper presented at an accident analysis conference.
These data contain a wealth of information for forensic investigators and traffic accident analysts and can help with a criminal investigation after a fatal traffic accident or an accident with injury.
With knowledge of how to decrypt the storage, the NFI carried out tests with a Tesla Model S so it could compare the logs with real-world data. It found that the vehicle logs were “very accurate,” with deviations less than 1 km/h (about 0.6 MPH).
The NSI also analyzed several accidents using the raw data it acquired. In one case, a Tesla on Autopilot collided with a car ahead that suddenly braked. Normally, if the Autopilot doesn’t brake in time, the driver is supposed to take over.
“In this case, the investigation showed that the driver did indeed intervene and also within the expected response time,” said researcher Aart Spek. “The fact that it turned out to be a collision was because the following distance [chosen by Autopilot] was too tight in the busy traffic situation. That makes it interesting, because who is responsible for the following distance: the car or the driver?”
It used to be possible to extract Autopilot data from Tesla EVs, but it’s now encrypted in recent models, the investigators said. Tesla encrypts data for good reason, they acknowledged, including protecting its own IP from other manufacturers and guarding a driver’s privacy. It also noted that the company does provide specific data to authorities and investigators if requested.
However, the team said that the extra data they extracted would allow for more detailed accident investigations, “especially into the role of driver assistance systems.” It added that it would be ideal to know if other manufacturers stored the same level of detail over long periods of time. “If we would know better which data car manufacturers all store, we can also make more targeted claims through the courts or the Public Prosecution Service,” said NFI investigator Frances Hoogendijk. “And ultimately that serves the interest of finding the truth after an accident.”