You probably already heard about this "self driving car" test... finding out if the car can tell the difference between a real road, and a picture of a road, in the middle of an actual road... just a flat surface it can not drive through. It's called the Wile E. Coyote Test

Tesla Autopilot technology relys on cameras,

Lexus RX-based prototype uses a lidar.

lidars scans the road ahead and detects the wall.

Autopilot’s cameras, on the other hand, rely on what they see, and in this case that’s a road.

But it is noteworthy just how effective the LiDAR system was shown to be in the video, as Tesla has very publicly decided to forgo these sensors in favor of relying entirely on computer vision. The reasoning for this varies depending on who you ask and when, but it usually boils down to LiDAR sensors costing too much, requiring more data processing to use, and ultimately serving as a crutch that slows down the development of computer vision. Tesla CEO Elon Musk has gone so far as to call LiDAR a “fool’s errand.”

https://www.thedrive.com/news/tesla-autopilot-fails-wile-e-coyote-test-drives-itself-into-picture-of-a-road

https://gizmodo.com/teslas-self-driving-fails-the-wile-e-coyote-test-2000577071