Soccer Robot System's Reasoning Research

Robot Soccer game is an international and high-tech oppositional activity that develops rapidly in recent years. It's a basic research project in the domain of artificial intelligence and robot, and also a great challenging and high-tech intensive project. Aim at the decision-making subsystem in a soccer robot system, the valid intelligent reasoning methods are discussed in this thesis.Firstly, we focus on introducing the mathematical model of the decision-making subsystem, the designing requirements and some system models, and analyze these models' disadvantage and advantage.Secondly, a simplified and multi-level hierarchical control structure of decision-making model for soccer robot is introduced. And the theoretical knowledge and applicable feasibility of three representative decision-making reasoning methods, namely fuzzy reasoning, expert system and neural network expert system, are discussed in detailed.Thirdly, according to the framework of decision-making subsystem, several improved methods are researched and designed: validity confirmation means of strategy based on fuzzy reasoning, method of role's allocation based on the situation of robots and ball, path-planning of robot based on the artificial potential field, and the PID control for wheel speed of the robot based on single neuron. These methods are simulated and contrasted with the self-attached method of the Robot Soccer platform, and are applied in the domestic and international games in 2004. Conclusions show these methods improve the reasoning competence of decision-making subsystem and the performance of the whole system.