Research Collaboration And Control Of The Robot Soccer System

A soccer robot system is the typical multi-agent system, where each robot in a team cooperates with its teammates under a real-time oppositional environment to fulfill specific objectives. The micro robot soccer tournament (MiroSot) system is a typical centralized control model based on central vision and decision-making subsystem. With the increasing of robots, the situation in the playground is becoming ever more complicated, and the cooperation among the teammates is getting ever more important problem.Firstly, the cooperation, the design and realization of agent in the Multi-agent system is studied, and a new real-time agent model is adopted. We proposed using the contract net model, which is usually used in the fields of distributed problem solving, to define the cooperative frame of the multi-robot. In this way, the coherence of robot' s actions is ensured, the cooperation among the robots is getting reinforced, and the design and realization of the system is becoming simple.Secondly, the role transformation strategies in a team are studied. A triangle transformation model is proposed. Further analysis based on Petri-net demonstrates the effectiveness of this strategy. It is efficient to realize the equilibrium distribution of robots in the playground, and it strengthens the attacking and defending in the local area.Thirdly, a new designed goalkeeper strategy is adopted. A new division method based on anomalous region is proposed, in which the determination of the region border is of enough evidence. And a new location method based on linear interpolation is presented in order to reduce the response time of the goalkeeper.