A Dirichlet Process Mixture of Robust Task Models for Scalable Lifelong Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore