Hybridizing evolutionary computation and reinforcement learning for the design of almost universal controllers for autonomous robots | Publicación