I think the brunt of the "Tesla Autopilot Deaths" problem is not the tech, but the marketing department (which has to include Elon Musk of course). This "kind of autonomous driving" is absolutely deadly, by definition. It's never going to work well. "The car will drive by itself, until it can't. So don't pay any attention, until you need to save your own life. We'll give you 2 seconds warning". I mean, really? I don't know why Tesla could possibly think this is a good idea. By extension, anyone who actually uses or trusts autopilot is basically forfeiting their life. I think it is cognitively harder to use autopilot and pay attention, than to simply drive safely.
As far as the tech goes, I am in agreement with Elon. I think LiDAR is a short term advantage, but the companies/stacks depending on LiDAR will reach a local maximum from which they won't recover without totally starting again from scratch. LiDAR is short range, low resolution, and extremely expensive. There is no reason that visible light (+ IR) CMOS cameras can't do better by an order of magnitude, at a lower cost.
As far as the tech goes, I am in agreement with Elon. I think LiDAR is a short term advantage, but the companies/stacks depending on LiDAR will reach a local maximum from which they won't recover without totally starting again from scratch. LiDAR is short range, low resolution, and extremely expensive. There is no reason that visible light (+ IR) CMOS cameras can't do better by an order of magnitude, at a lower cost.