It’s called consensus. Have three sensors and each get a vote. Typically these sensors are the same and thus can detect a failure or incorrect reading of one. This idea is used in IT around data backups and RAID configurations as well as aviation. And … I personally would just favor the radar. If vision says go and radar says stop… stop and avoid hitting that firetruck parked on the highway. Or that motorcyclist. Or any other bizarre vision-only, fatal crashes that this system has wrought.

Also humans can hear things. So, not just vision.

Yeah… But these are multi-ton vehicles and when they crash people die. Unlike when your computer crashes.