The National Transportation Safety Board has published its final report on a fatal 2016 crash involving a Tesla Model S 70D which had its Autopilot engaged.

On the afternoon of May 7, 2016, a 2015 Tesla Model S 70D was being driven by 40-year old Joshua Brown on US Highway 27A in Florida. The car struck a semi which was making a left turn across the two oncoming lanes of traffic, shearing off the Tesla’s roof before coming to rest in the front yard of a private residence. Brown was killed; the truck driver was not injured.

In its report, the NTSB determined that the probable cause of the crash was the truck driver’s failure to yield the right of way to the Tesla, combined with the Tesla driver’s inattention due to over-reliance on vehicle automation. This second factor, asserts the NTSB, resulted in Brown’s lack of reaction to the presence of the truck.

READ MORE: Fatal Tesla Crash Involved Speed and Autopilot

Speaking directly to the use of this specific autonomous driving feature and how it is designed, the report stated that the operational design of Autopilot, which permitted prolonged disengagement from the driving task, contributed to the driver’s overreliance on vehicle automation. It went on to say that the design permitted use of the of automation inconsistent with guidance and warnings from the manufacturer.

The investigation focused on several different safety issues, including surrogate means of determining the automated vehicle driver’s level of engagement. The NTSB recommended, quite rightly, that manufacturers need to develop better ways of monitoring the attention level of a driver who is employing autonomous technology. Pressure-sensitive steering wheel sensors can theoretically be defeated by taping a water bottle to the wheel, for example.

From the report:

Although the operational design of the Tesla Autopilot requires an attentive driver as an integral system element, the Autopilot on the Williston crash vehicle allowed the driver to operate in the automated control mode for almost 6 minutes, during which the system did not detect the driver’s hands on the steering wheel. In fact, of the 37 minutes in which the Tesla driver operated the vehicle in the automated control mode during the crash trip, the system detected his hands on the steering wheel for only 25 seconds.



Investigators found no indication in recorded vehicle data that the Tesla driver attempted to take any action (by braking or steering) to prevent crashing into the semitrailer or that he was even aware of the impending crash.


Better methods of monitoring driver engagement should be deployed, the report said, such as inward-facing cameras which track driver eye movement. GM’s new Super Cruise feature in the Cadillac CT6 employs this method, while Toyota has installed an eye-tracking system in some Lexus vehicles. Volvo has announced plans to use this type of technology in its Driver State Estimation system as well.

Christopher Hart, board member of the NTSB, added a statement to the report, drawing parallels between the current environment of emerging vehicle autonomy and the airline industry. GM’s new Super Cruise feature employs a version of this type of monitoring.

Mr. Hart goes on to take the name “Autopilot” itself to task, saying that while trained pilots know they still play a crucial role in controlling the plane even when autopilot is engaged, John Q. Public may erroneously conclude from Tesla’s naming of the feature that they need not pay any attention to the task of driving because all duties are being handled by autopilot.

The full report, spanning 53 pages, can be found on the NTSB website.