The accidents involving large stationary objects were due to improper usage of the system, users still need to pay attention and take control. As you said, there aren't any truly non-commercial self-driving cars yet, so how can you say the solution is faulty without using lidar? The current solution simply wasn't programmed to do it as users are expected to follow the instructions, doesn't mean it can't be achieved. Tesla's have both radar and vision and with hw3.0 (FSD HW), the software hasn't even caught up to the HW limitations.CanadianLurker wrote: ↑ There aren’t any non-commercial self-driving vehicles currently available at all.
That said, most consumer vehicles that have adaptive cruise control use a radar-based system. Not sure if the terminology is radar or lidar but there are millions of vehicles on the road with adaptive cruise control that use radar/lidar to detect objects ahead and then change speed (or brake) accordingly.
Tesla/Musk steadfastly refuse to use such systems and so Tesla’s all rely on cameras only. There have now been several Tesla collisions involving large stationary objects blocking the raid that the cameras have failed to identify as a problem.
Personally, I think that the best approach would be to use both systems. Use the cameras as the main system and radar/lidar as a supplemental system for an added margin of safety. But given Musk’s stubbornness it isn’t likely to happen unless there are more collisions that finally force his hand.
While I'm sure Lidar would be useful, it's simply not economical IMO to pass it onto the end consumer when EV's are still expensive in comparison to ICE cars. Musk is not against LIDAR technology either, it's used on Crew Dragon to dock onto the ISS. It has its place and if autonomous driving level 4-5 can be achieved without it, then the better it is for the end consumer.