Robotic inverse kinetics learning.

[iplayfast]

I've an idea: A robot with n servo's needs to learn to walk. I know this can be done using gazebo and other software, but I think PySR might be a shortcut. First train to stand. This should be able to figure out quickly as it is a simple PID algorithm. The parameter values needed are what is calculated by PySR. The interesting thing is that we could do real life sampling of the tests, and use the results to adjust the goal. Once standing is figured out, then move to recovering from falling, then walking.

My idea is not well thought out, but I thought I would throw it out for discussion.