HYBRID INTELLIGENT LEARNING ALGORITHM BASED ON MULTI-AGENT AND ITS APPLICATION TO ROBOT SOCCER
ZHANG Shu-jun1, MENG Qing-chun1,2, SONG Chang-hong1, LI Zhan-bin1, ZHANG Wen1
1. Computer Science Department, Ocean University of China, 266071; 2. State Key Lab of Intelligent Technology and Systems, Tsinghua University, Beijing 10004, China
Abstract:This paper presents a hybrid intelligent learning algorithm based on multi-agent, and it combines individual learning with group learning effectively. The specific application and experimental results in RoboCup simulation system show the feasibility and effectiveness of this algorithm.
[1] Zhuang Xiaodong, Meng Qingchun, Wei Tianbin. Robot path planning in dynamic environment based on reinforcement learning [J].Journal of Harbin Institute of Technology,2001 ,8(3): 253-255. [2] 王立春,李红兵,陈世福.AODE中的多Agent协商与学习[J].模式识别与人工智能,2001,14(3):347-351. [3] 李实,徐旭明,叶榛,孙增圻.国际机器人足球比赛及其相关技术[J].机器人,2000,22(5):420-426. [4] Peter Stone. Layered learning in multi-agent systems [D].Pittsburgh:Carnegie Mellon University: 98.