MAMBPO: Sample-efficient multi-robot reinforcement learning using learned world models

Open in new window