[....] For City Streets, the primary function that is added is finally the car being able to make left and right turns on intersections.
For "City Streets" do you mean "NOA for city streets", as tested by the small band of FSD beta testers,
or "Autosteer on City Streets", which exists now for everyone with FSD intrepid enough?
What I'm getting at (as I've said elsewhere), why do many assume that Autosteer on City Streets
will do hard turns at intersections? It's easily possible that autosteer in town will act just as it
does on the highway and allow for driving straight with lane changing, etc., but with the added
functionality of working on streets without lane markings, guiding thru narrow streets (on curves, too)
including parked cars, nudging around cars with open doors, driving around double-parked cars when
it can, going thru intersections w/o confirmation -- all much welcomed.
But the NOA-on-city-streets folk are testing 90-degree intersection turns not by putting on
a turn signal to have autosteer take the next right or left, but by using what is essentially NOA.
From the existing videos, the public seems not quite ready for such drastic actions. If high-radius
turns are known to be tougher from a safety perspective, then I see no reason why Tesla won't
be conservative with "Autosteer on City Streets" when it's rolled out to a wide audience.
Even the more limited functionally of *everything but intersections* requires major changes
beyond current autosteer, i.e. multi-camera stitching to the vector-space domain, totally revamped
NNs, a completely different path planner, etc. That's huge, and we don't even have Tesla Vision
for radar cars yet. Safety first, which to me implies gradualism with new features.