But if you really think about it, flipping rocket is a trivial toy problem in simulation. But useful humanoid robot is very hard even in simulation. Just look at some of the SOTA control theory research. Their simulated robots looks stupid.
Flipping that rocket and getting NASA contracts to go actually do it against Boeing/Lockheed/etc is not a toy problem.
I don’t want to say it would never have happened without Spacex, but probably it would not have happened in our lifetime.
Human-level robotics, on the other hand, is not a problem of missing incentive. For all I know about the field (as a casual observer), we are still several breakthroughs away from achieving this goal.
We have actuators, sensors, and computers to build humanoid robots, isn’t it “just” a software problem like flipping rockets was?
There's a difference between not being able to do something, and not having a reason to.
Enough.