Transfer Reinforcement Learning Across Homotopy Classes | IEEE Journals & Magazine | IEEE Xplore