Feature Control as Intrinsic Motivation for Hierarchical Reinforcement Learning | Read Paper on Bytez