Raw LLM Responses

First of all - It is still a Beta feature in development. As it is at the moment equivalent to "driver assistance", the driver is responsible for everything. As there is no perfect computer decision, there are many more worse human decisions. You can argue that there are good drivers... but how many are actually "good"? Even airplanes need two pilots when operating in "autopilot mode"... in a nearly empty sky. Radar was removed because it became an obstacle. You can't rely on only radar... And when two (or more) sensors disagree with each other, the measurement needs to be done again. That takes up a lot of critical time to make a decision. A radar can take you up to a point in the "self driving" category... After that it becomes a barrier. Removing the radar removes miles of cables and additional equipment. Which in turn removes many failure points. Computer vision has much more potential at "seeing", understanding, measuring and making a decision faster. It already scores 98% in crash tests and is just in the beginning of development. LiDar is indeed a fool's errand... At least for passenger vehicles. It has lots of industrial applications, but in all cases - in a localized scale. This is the limiting factor - it needs pre-build information (map), needs expensive equipment, need much more additional power (less range), needs faster heavier and louder computers, needs heavy and bulky sensors. LiDar does NOT have the ability to work on unknown road or in fast changing road condition. Does NOT have the ability to worn in rain, fog, snow, smog or even wind. So at the end of the day it's a "cost saving"... for the customer.