運用同化與調適於足球代理人的合作學習

Applying Assimilation and Accommodation for Cooperative Learning of RoboCup Agent

游怡婷(Yi-Ting Yu)

指導教授 ： 郭忠義

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2009年)

https://doi.org/10.6841/NTUT.2009.00185

摘要

適應性學習是在多代理系統中改進收斂速度與學習品質的關鍵能力。一個靈活且具合作的機制需要的代理人彼此合作地學習。本論文整合三個適應性學習方法，使機器代理人快速且有效率的學習。強效式學習(Reinforcement Learning) Q-Learning方法用於學習動態多變的足球競賽的策略；當外來知識與機器代理人擁有的知識衝突時，使用同化(Assimilation)技術調整代理人知識，而外來知識為新知識時，則運用調適(Accommodation)技術接收；最後歸納為模糊規則提供機器代理人快速推論競賽所需動作。我們使用RoboCup模擬平台來說明所提出的方法。

關鍵字

同化 ； 調適 ； 強效性學習 ； 模糊推論

並列摘要

Adapting learning is the essential ability to improve the convergence rate and learning quality in the multi-agent system. A cooperative mechanism needs learning cooperatively between agents. This research integrates three adapting learning method to make agent learns efficiently. Reinforcement Learning is used to learn the strategy of the dynamic soccer competition. If there is a conflict between the external knowledge and the agent’s own knowledge, we use the Assimilation technology to adjust the agent’s knowledge. And if the external knowledge is a new knowledge, we use the Accommodation technology to receive. Finally, the fuzzy rule provides the agent with the action which the competition needs. We use the RoboCup simulator to explain our research.

並列關鍵字

RoboCup ； Assimilation ； Accommodation ； Q-Learning ； Fuzzy

參考文獻

[2] L. Vig, J. A. Adams. Multi-robot coalition formation, IEEE Transactions on Robotics, Vol. 22, Issue 4, pp. 637-649, 2006.

[3] A. Rao, and M. Georgeff, “BDI Agents: From Theory to Practice”, International Conference on Multi-Agent Systems, San Francisco, USA, 1995.

[4] Yilu Zhang; Juyang Weng. Task Transfer by a Developmental Robot. IEEE Transactions on Evolutionary Computation, Vol. 11(2), pp. 226 -248, 2007.

[5] R. S. T. Lee, J.N. K. Liu, “iJADE Web-Miner: An Intelligent Agent Framework for Internet Shopping”, IEEE Transactions on knowledge and Data Engineering, Vol. 16, pp. 461-473, 2004.

[6] T. Nakashima, M. Takatani, M. Udo, H. Ishibuchi, “An evolutionary approach for strategy learning in RoboCup soccer Systems”, IEEE International Conference on Man and Cybernetics, Vol. 2, pp. 2023-2028, 2004.