This is not obvious. We should not discount the data collection and training infrastructure that Tesla has created. It’s very possible that their massive fleet of robots on wheels gives them a significant leg up.
Tesla has a mediocre "self"-driving car, and a robot primarily designed for looks which seems to be on a level achievable by the average university student team.
How is training data primarily about things like recognizing traffic signs going to be of any use when the hard problems to solve are stuff like picking up a box without crushing it?
How is training data primarily about things like recognizing traffic signs going to be of any use when the hard problems to solve are stuff like picking up a box without crushing it?