Don't blame the vision! The problem of Tesla's "phantom braking" is caused by the Radar malfunction!
Tesla's AI and autonomous driving vision director, Andrej Karpathy, explained this issue at an autonomous driving seminar last year. Due to Radar not having sufficient vertical resolution to distinguish whether the object ahead is a bridge or a car, it created a stationary target and triggered the braking, while the vision itself can generate depth and speed, but this accuracy was obscured by the associated Radar.
As a result, Tesla decisively abandoned Radar and turned to the development of pure vision autonomous driving, and has already solved the "phantom braking" issue.
Therefore, Andrej Karpathy ultimately believes that a single type of perception system is more efficient than a combination of Radar and vision; having one more Sensor only adds another layer of trouble.
Now, the pure vision FSD beta that Tesla is using has become increasingly mature, and there is even a possibility of achieving fully autonomous driving this year, so the autonomous driving solution that combines Lidar and vision needs to catch up.
$Tesla (TSLA.US)$ $Tesla Ecosystem (LIST2436.US)$