Abstract: Reinforcement learning is an approach that an agent can learn its behaviors through trial-and-error interaction with a dynamic environment.It has been an important branch of machine learning for its self-learning and online learning capabilities. But reinforcement learning is bedeviled by the curse of dimensionality.Recently，Hierarchical Reinforcement Learning（HRL） has made great progresses to combat the curse of dimensionality.And the HRL approaches have been being applied to multi-agent system.The recent advances in HRL are surveyed in this paper.Then，some open problems are discussed.Finally，the HRL prospects are shown.

Key words: Hierarchical Reinforcement Learning, multi-agent system, curse of dimensionality

摘要： 强化学习通过试错与环境交互获得策略的改进，其自学习和在线学习的特点使其成为机器学习研究的一个重要分支。但强化学习方法一直被维数灾难所困扰。近年来，分层强化学习方法在解决维数灾问题中取得了显著成果，并逐渐开始向多智能体系统推广，论文归纳分析这一领域目前的研究进展，并对迫切需要解决的一些问题和进一步的发展趋势作出探讨和展望。

关键词: 分层强化学习, 多智能体系统, 维数灾难