Cameras instead of LiDAR for safety = fail

I found the video to be very interesting and the results were not surprising with one exception: the rain test. I was surprised the lidar was able to make out the mannequin through all the water--especially after seeing all the posts about rain affecting the lidar sensor on the Air. However, in this case, the sensor itself stayed dry for most of the test which probably improved performance. In a real world scenario, the lidar sensor cover would probably have water droplets on it and may not function as well.
 
I don’t understand why lucid doesn’t turn on more fsd capabilities given they have both lidar and camera.
 
Beat to the punch. Excellent video that anyone could understand.
 
Experiments to fool the camera with murals are interesting. There's also plenty of history of human drivers being fooled. I seem to recall a trompe l'oeil painting of a tunnel on the side of a building, which was quickly repainted after cars started ramming into the wall. Personally, I'm hoping that autonomous driving is ultimately much much safer than humans driving, and it seems likely that super-human perception through more sensors than cameras will help achieve that.
 
Back
Top