I do wonder if he's doomed FSD/robotaxi by being stubborn about lidar. The self-driving problem is unlike most other ML problems due to its mission critical nat

I do wonder if he's doomed FSD/robotaxi by being stubborn about lidar. The self-driving problem is unlike most other ML problems due to its mission critical nature. Having 2 modalities that can be cross-checked will reduce edge case hallucations that cause crashes. Like, instead of probability p that you don't see the truck in the middle of the road, you reduce it to nearly p^2. This non-linear reduction in faults that comes with cross-checking modalities is not intuitive. I never bought Musk's justification: "Lidar is not how humans do it". Why would we intentionally handicap ourselves to the tools available to biological evolution? Seeing how hard it's been for Cruise and Waymo, who use lidar, and only operate in a small area, I am not expecting a functioning Tesla robotaxi that regulators will allow on the road anytime soon.