Taking R&D out of the picture Vision+LIDAR sensors would be more expensive than Vision alone but still Musk always believed and is adamant that perfecting vision-based navigation is essential for the development of automation in general. The fork in the road and all that. It's not only about driving. I don't think saving money was a big reason.
Perfect vision-based navigation is still going to hit obstacles. It's a "first principle" that active sensors have more potential to detect things than passive ones.
Have you ever heard a sci-fi where they turn off active sensors to avoid detection and the Captain says "actually just leave them off for good they're pointless". No, they are always 'flying in the dark' because everybody knows active sensors are better.
Telsa used a cheap $20 camera sensor instead of a $50 one that was across the board better. Of course dropping lidar was about money.
Even Submarines have to use their sonar at some point just to see what is around them and correct for minor drift in course. Most of the time they use passive sonar (i.e. listening to the ambient noise) to avoid revealing their location, but even with gyroscopic inertial navigation assistance, tiny errors accumulate over time, and they have to send out a sonar pulse to see what is around them in the water.
They also periodically surface to attain a gps signal and receive comms, since it's pretty hard to penetrate 1000 feet of water with a radio signal.
Intended audience for the video is 5 year olds, cartoon voice, staged and manipulated. It's painful to watch more than 15 seconds of it.
But it's basically correct. Lidar is always going to be a better technology for avoiding car crashes.
Tesla went vision-only to save money, not because it's better.
Taking R&D out of the picture Vision+LIDAR sensors would be more expensive than Vision alone but still Musk always believed and is adamant that perfecting vision-based navigation is essential for the development of automation in general. The fork in the road and all that. It's not only about driving. I don't think saving money was a big reason.
Perfect vision-based navigation is still going to hit obstacles. It's a "first principle" that active sensors have more potential to detect things than passive ones.
Have you ever heard a sci-fi where they turn off active sensors to avoid detection and the Captain says "actually just leave them off for good they're pointless". No, they are always 'flying in the dark' because everybody knows active sensors are better.
Telsa used a cheap $20 camera sensor instead of a $50 one that was across the board better. Of course dropping lidar was about money.
Even Submarines have to use their sonar at some point just to see what is around them and correct for minor drift in course. Most of the time they use passive sonar (i.e. listening to the ambient noise) to avoid revealing their location, but even with gyroscopic inertial navigation assistance, tiny errors accumulate over time, and they have to send out a sonar pulse to see what is around them in the water.
They also periodically surface to attain a gps signal and receive comms, since it's pretty hard to penetrate 1000 feet of water with a radio signal.