The Research Of Coordination Strategy Based On MAS In Robotic Soccer Simulation

In this thesis, the coordination strategy of MAS applied in the field of RoboCup simulation is proposed. During the match, each agent obtains skills which include dash with ball, pass ball and so on . These agents implement their team positioning, formations and behavior modes through the real time strategy of task decomposition, multi-layer learning and dynamic role, in order to achieve the goal of the coordination.The dissertation was described as follow:At first, the theory of MAS and its application are introduced, the survey of various outstanding research institutions at home and abroad in this field is also shown, and the research plan of ours is established according to the survey. Then the various characteristics of the soccer server are described, which includes the sense model, the movement model, the action model, the heterogenous agent and the match mode, moreover the modifications of the soccer server between the new version and the old version are also introduced.Secondly, the architecture of the agent is described. In the system analysis of the agent, the agent system is composed by models in different layers, as well as various models are associated with various threads, then the agent system is described in detail by the UML class diagram and the sequence diagram in the view of object oriented. In the system design of the agent, the coordination strategy of multi thread based on PV operation is proposed according to the system analysis, so that the agent would be suitable for the simulation environment.Thirdly, the strategies of the visual model, communication and the information fusion are also proposed, and the experiment has shown that the optimized strategy is effective. Then the macro strategy which formally describes the relationship between the team strategy and the agent behavior is proposed, and the agent behavior is controlled by the static and the dynamic method in several aspects, the experiment shows the features of the different solutions. At last, the innovation of the paper is proposed, the result of the RoboCup China Open2007 is introduced and the shortcomings of our team in the competition are concluded, then the development direction of the research in future is described.