Competition and Cooperation Strategy in Multi-agent System
Li Mao · Journal of Nanjing University of Science and Technology · 2005
RoboCup(the Robot World Cup) Simulation is used as a platform to study multi-agent competition and cooperation strategy in a complex environment.Firstly,the approach of opponent modeling is expounded,which takes corresponding strategy by analyzing the weakness of opponent.The approach enables the agents to make more effective independent decision and makes the actions highly pertinent and objective.Secondly,the information sharing among agents is analyzed.The cooperation among agents is improved by compressed coding and selective information sharing.Finally,local cooperating tactic problems are analyzed.The approach implements the local cooperation among agents through designing the tactic,and adjusts the tactic according to the change of the field condition.The approaches were applied to the China Robot Competition 2004 and gained the champion.That proves the method is efficient.