Obviously there are fundamental problems.
One aspect of animal intelligence Tesla could seek to understand is that organic brains are continually in the future. Future possibilities (determined by past experience) are always primed, always ready to go, always changing according to current input. Minimal reaction time is life saving. Possible reactions are being primed with each atom of sensory input as it arrives – given the atoms of sensory input just received over the past X seconds. It’s a continually adjusting fused flux of possible reaction.
Until Tesla works out how to embody basic such life-saving aspects of animal intelligence in its machine, it ought to stop offering causal systems that are much-touted, relentlessly over-hyped and at least in part, death traps. At least it ought to consider that such offerings of partial or full human death traps are not the soundest possible business strategy going forward.
It could think about winding back its Autopilot offering, and concentrating on trying to understand the basic principles of human-like intelligence. Why would a competent enterprise go forward into autonomous vehicle control system production and sale without doing this?