@edgecase411 As it’s executing the turn the yellow line on the right dissapears and the perception system thinks only the yellow line on the left exists. So the planning system thinks that lane is actually valid to drive in.
@cvalente28@MuldoonMartin At the same time, I do agree they should start educating and advertising. The product already has amazing word of mouth reputation. Further education can only help and doesn’t need to cost too much.
@cvalente28@MuldoonMartin I think you’re not considering the other side of the coin….the short and distort campaign from ‘17-‘19 nearly bankrupted the company and caused a great deal of pain. They manage the company for the long-term, full stop. Shareholders have not been forgotten, traders have.
@wintonARK Do you think diffusion is needed for robotaxi? Seems like the current approach isn’t quite getting there for Tesla. Some of it still might be perception but the twitchiness of the planner is a big problem for testers
@JPSartre_NoExit Any interest to discuss how deliveries & insured units data can diverge - a lot of negative sentiment around recent insured units data in china
@wholemars@elonmusk The ITC can be claimed by the owner of the megapack. If Tesla sells these to independent installers they won’t be claiming the credit.
@WOLF_Financial@wholemars do you think Elon has changed his mind about needing neural networks in the planner for FSD? IMO the planner seems to be causing a lot of the common issues people have (lane selection, random breaking, hesitancy in decision making) but I could be wrong
@davelee@jamesdouma Questions about the FSD planner… how much of it he thinks is NN based? will it improve exponentially to handle all weird corner cases humans are able to interpret?
Contrary to public opinion, legacy auto companies are at a massive disadvantage to Tesla in producing electric cars. The ICE has been their core competency - not batteries, electric power trains, or even manufacturing! Tesla’s vertical integration in manufacturing is unrivaled.