Multi-Agent Deep Deterministic Policy Gradient (MADDPG)