关键词: 增强学习, 半自治agent, profit-sharing, Q-learning

Abstract: We exert the profit-sharing reinforcement learning method into the semi-autonomous agent system，and compare it with the other reinforce learning method——Q-learning.Profit-sharing method is more robust and fit for the dynamic environment which includes many uncertain factors，especially in the partial MDPs（Markov Decision Processes） environment.Facing the semi -autonomous property of the agent，we propose an improving learning method of profit-sharing in the semi-autonomous agent system and test it in a combat simulation environment that finds the safety hidden space in battlefield.At last we contract and analyze these methods to the others.

Key words: reinforcement learning, semi-autonomous agent, profit-sharing, Q-learning