jyalpert
Member
This is what I don't fully understand about Tesla's autonomy program, and none of the analysts asked about it during autonomy day. What's the strategy?LDA and ELDA are disguised as safety features, but rather two of the many building blocks of FSD. I think Tesla's definition of feature complete means to demonstrate these mini-features. How many to surface? I don't know.
If their architecture is right and all the necessary building blocks are exhausted and designed correctly, FSD will eventually work. However, if most of the blocks are not working reliably 100%, the whole FSD would fail easily.
Andrej described a painstaking, piecemeal approach of essentially making a long list of things that happen while you're driving, and trying to create individual point solutions for each one of those things. He cites the cut-in as an example. I'd guess lane departure is another example.
Somewhere on that list of things, they'll draw a line and say "now we have feature complete FSD". It's a good strategy for them, as they can release each point solution independently and call it an additional driver aid.
But I'm not clear on where that line is, where they are vis a vis that line, and whether that line is "good enough" to actually create a car with no steering wheel or pedals.
This, plus many other red flags and contradictions in the autonomy day presentation, have left me more confused than before.