A jarring dashcam recording showcases how a Tesla Model X, running on autopilot, rammed into a stationary police vehicle on February 27, 2021, on a highway near Montgomery County, Texas.
The footage, obtained by the Wall Street Journal, showcases a Tesla Model X driving at 54 miles per hour on the rightmost lane of a highway. After passing multiple parked emergency vehicles, a police SUV with its lights flashing appears stationary in the center of the rightmost lane, which the Tesla fails to detect.
There are at least six vehicles on the scene before the crash, many of which are flashing lights. An attentive driver would have likely recognized them and slowed down or changed lanes. The EV then rams into one of the police SUVs, injuring five officers and hospitalizing the subject of the original traffic stop, as per the video.
Autonomous driving experts WSJ spoke to suggested that the police vehicle’s strobing lights might have created a hazy image that the car’s software likely did not recognize. The system detected an obstacle in its path 37 yards and 2.5 seconds before the crash, which is obviously too short of a distance for a safe stop from 54 mph.
The police revealed that the driver was intoxicated, and the Model X alerted him 150 times in a 45-minute span to apply slight force to the steering wheel, to which he complied. However, the intoxication possibly explains why he needed more than three alerts per minute to keep his hands on the steering wheel.
The National Highway Traffic Safety Administration (NHTSA) is investigating 16 such instances of Teslas crashing into emergency vehicles to determine the cause of the accidents.
In one document, the NHTSA mentioned that the forward collision warning (FCW) systems were activated immediately before the impact while the autonomous emergency braking (AEB) intervened in half of the collisions. NHTSA also mentioned that “on average in these crashes, the autopilot aborted vehicle control less than one second prior to the first impact.”
The 2019 Model X uses cameras and radar to detect objects in its environment. Tesla has mentioned several times on its website that autonomous features are to be used under active driver supervision.
The Austin-based carmaker is reportedly fusing artificial intelligence capabilities with its self-driving software, with an aim of reaching SAE Level 4 or 5 autonomy by the end of this year. Currently, Tesla vehicles are capable of Level 2 autonomy.