Multi-Agent Coordination: A Reinforcement Learning Approach