Follow a clear path through difficult ideas.
Move from policy gradients and search to practical robot learning with demonstrations and human feedback.
