Tesla CEO Elon Musk (Elon Musk) reportedly announced nearly two years ago that Tesla would stop installing radar sensors in its cars. Data shows that the number of subsequent accidents and near-misses of Tesla cars has increased.
Interviews with dozens of former Tesla employees, test drivers and other experts show that after the 2021 upgrade, Tesla cars driving through Autopilot or FSD features are braking more often because of non-existent obstacles, misidentifying street signs, and having difficulty identifying emergency vehicles.
Some sources say the increasing number of false braking cases is related to Tesla’s decision to remove radar sensors from its vehicles. Data from the National Highway Traffic Safety Administration (NHTSA), which is investigating the issue, shows that the agency has received hundreds of complaints about false braking in the past nine months. Last year, more than 750 Tesla owners complained that their cars suddenly and inexplicably braked while driving.
Meanwhile, NHTSA also stepped up its investigation of Tesla’s Autopilot feature in 2022 after more than a dozen accidents in which Tesla cars crashed into emergency vehicles. NHTSA said the driver-assist feature had difficulty identifying parked vehicles.
Musk initially announced that Tesla would stop having radar sensors in its cars starting in 2021. Some engineers were reportedly “taken aback” by the statement and contacted a former Tesla executive in hopes of convincing Musk to change his decision. Musk has also said in the past that he wants Tesla’s FSD and Autopilot software to simulate the senses of human drivers through cameras rather than radar.
Currently, all Tesla cars are equipped with Autopilot driver assistance. Users can also pay a one-time $15,000 or $199 a month to enable FSD, which helps cars recognize stop signs and traffic lights, automatically adjust lanes and park themselves. But Tesla says neither Autopilot nor FSD can replace a licensed driver.
Until 2021, Tesla cars will use radar sensors in addition to cameras to identify obstacles. Currently, Tesla relies on eight cameras and Autopilot image taggers to train cars to react to their environment. Tesla employees also tag videos taken by the car’s cameras to train the software to recognize and respond to different obstacles.
Other self-driving sensors, such as LIDAR, are also used by Tesla’s competitors. These vehicles use LIDAR to digitally map the environment and avoid errors, even if the onboard cameras are obscured by external obstacles, such as rain, snow and fog. Musk, however, has previously said that LIDAR is too costly and therefore “doomed to fail.”
Since 2016, Musk has been promising that Tesla will soon launch a truly self-driving car, but experts are not optimistic.
Earlier this year, several experts said that the Tesla FSD is still a long way from self-driving. In February, Tesla released an OTA software update to 362,000 vehicles to address a problem with the FSD, which the NHTSA said at the time could cause cars to “behave unsafely at intersections.
A Tesla spokesperson did not comment. In a voluntarily released vehicle safety report, Tesla said it had the lowest overall probability of injury among all vehicles tested by the U.S. government’s New Car Evaluation Program.
In January, Tesla also said that in the third quarter of 2022, the probability of a Tesla accident was already as low as one accident every 6.26 million miles (about 10.074 million kilometres).