Feature Control as Intrinsic Motivation for Hierarchical Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore