Exactly. "all other companies use radar." Do people think Tesla is so far ahead by doing everything every other company is doing? Wasn't there a tweet and an article that outlined how to there were conflicts between information gathered by LIDAR and information from the cameras that were making AP/FSD more complicated and that going to a camera-only system would be more accurate and efficient in the long run? Do people really think that with year-long wait times and 30+% margins on every vehicle, that they deleted the LIDAR just to save a couple hundred bucks a car? Come on, people. It's not "there was a feature and it was deleted." No feature was lost.
I've driven Teslas with LIDAR and most recently Teslas with Vision only. There is no noticeable difference to me.
Tesla does a lot of things right by not following the charted path. But that does not mean that they have to challenge _everything_. In fact, challenging everything is lack of wisdom, typical for a lot of startups. Understanding what to challenge and what to use is a sign of mature leadership. It is time for Tesla to grow up from the startup past.
There are many, many examples where they challenged existing knowledge without much understanding just to come back around and accept it. V11 fiasco was probably the latest example.
Currently, they have the market because of the advanced drivetrain and charging network. The rest of the car is pretty average, maybe even below average (would you buy an ICE Tesla?). The dealership network is mediocre (aside from the mobile units).
Tesla can allow to mess up maybe a few more times and then those mistakes will become very costly. There are cars that at the same price point are superior in terms of ergonomics, design, support and quality. It is just a matter of time for them to figure out the drivetrain. Tesla is so much focused on scaling and distractions like AI, FSD, monkeying with software, that, aside from the batteries, they have not done much innovation in the drivetrain. The difference between Raven and Plaid is purely cosmetic, with some marketing stints like half-wheel dropped in. In its essence, the car is the same plus one more motor. Cybertruck is delayed again; Roadster is nowhere near production.
Specifically about so called “Tesla Vision”. The arguments for it is a pure BS (don’t trust me - Google experts opinions). Visual ranging is inferior to radar ranging by definition. It is slower because it calculates parallax (angle difference) between images vs. simple S=V*T. It is also less accurate, especially at closer distances when the images are vastly different. No wonder visual ranging is limited to only situations where active radar cannot be used. That inaccuracy and lower speed are the reasons why the following distance is increased and max AP speed is reduced.
Removing a perfectly functioning sensor because it disagrees with the other is another sign that Tesla just does not get it and FSD will continue to be “two years away”. Driving is one of those problems with long tail end scenarios and you have to solve for 100% of them; 99% will not cut it. Discrepancies are solved by adding more sensors to cover more scenarios, not by removing them. Yes, radar may be worse than visual in 20% of the cases but it will be invaluable in that 1% that you really need it.
Removing the radar was a measure to circumvent the supply chain problems and they used to opportunity to cut cost. Have in mind that the cost of the sensor itself is minimal; supporting the software is where the real money is.
Even though I also lost trust in Tesla I love what they have done and I still have a sliver of hope that they will concentrate on what they do best - making great EV - and stop those distractions.