Even though Tesla stopped using radar in their vehicles last May, they plan to add a new radar product come mid-January, as stated in documents submitted to the Federal Communications Commission.

Tesla began removing radar from its vehicles last May. In October, the company removed its 12 ultrasonic sensors from Model 3 and Model Y vehicles built for North America, Europe, the Middle East and Taiwan. Ultrasonic sensors measure distance via ultrasonic waves and are used as proximity sensors to support anti-collision safety systems, particularly in parking use cases.

Now it appears radar is going back. Tesla had originally filed with the FCC to use the new radar - which is described in filings as "76-77 GHz Automotive Radar" - in its vehicles back in June.

"From the frequency of operation (76-77GHz) as well as the mechanical design of the sensor from Tesla's FCC filing, it appears that this radar would be utilized in ADAS applications," Steven Hong, VP and general manager of radar technology at semiconductor company Ambarella, told TechCrunch.

He noted that while the performance of this 'edge' radar sensor will be limited, it's a positive development that Tesla is looking to add radar to its perception stack for safety-critical, robust performance.

It's not yet clear which models will get the new radar. The type of radar Tesla intends to market next year is of a frequency that's allocated by the FCC for ADAS use cases, according to Ram Machness, chief business officer at Arbe Robotics, which produces ultra-high-resolution 4D imaging radar.

The luxury EV-maker has long claimed it could reach full autonomy through a "vision only" approach that shuns other sensors like lidar and radar in favor of cameras and a deep neural network that quickly processes a vehicle's surroundings and responds in real time. Tesla CEO Elon Musk previously promised to "solve" full self-driving by the end of this year (he's also promised Tesla would get there every year for roughly nine years now). He has recently admitted the problem will take longer to solve. 

And perhaps, as every other autonomous vehicle technologist says, it's not actually achievable yet through cameras alone.

Source: Techcrunch