Based on Reinforcement Learning

摘要：

This paper introduces the current study of reinforcement learning and describes a system which can make agent use the marriage of Q_learning and neural network to form team and avoid collision .In the multiagent system ,there are many initial problems such as big search space ,difficulty in knowing other agents, and the low robust of the system.. Using the way which described in the paper to make team formation of mulitiagent can solve these problems very well. This way has the good expansibility.

关键词： reinforcement learning multiagent team formation Q-learning neural network

作者: Zhang Rubo Wang Xingce

作者单位: The science and technology of computer college, Harbin Engineering University

会议类型: 国际会议

会议名称: 2007国际导航制导控制学术会议

会议地点: 北京

会议语种:英文

页码: 217-220

在线出版日期: 2007-11-04（万方平台首次上网日期，不代表论文的发表时间）