Research On Attack Strategy Of Micro Soccer Robot

Robot soccer is a question for discussion in the field of artificial intelligence androbot. It's a challenging and high technological problem. The robots are conducted bystrategy subsystem, so the integrality and advancement of strategy system are themost important for soccer robot system.The attack strategy subsystem discussed in this paper is a part of all strategysystem. The research objects are micro vision-based soccer robots. The main missionincludes designing a series of basic behaviors according to the characteristic of robots,which could be assembled to arrange more advanced acts, planning an optimal attackpath for each robot, and introducing the multi-robot cooperative mechanism to makerobots in the field complete the attack mission by collaboration.Firstly, this paper presents the introduction of the type of soccer robot andanalysis of the strategy system, including the tasks, characters of strategy system andthe model of it. I also detailedly discuss the idea of hierarchical structure used instrategy system. Secondly, the motion planning problems of soccer robots arediscussed. Path planning method based on Rapidly-exploring Random Tree is theanalytical emphasis. Furthermore, in order to forecast the trajectory of the ball,Kalman filter is put forward, which can estimate the possible time when the robot canreach the moving ball and the point where the robot can kick it. Thirdly, the methodthrough which the attack behaviors of robots are designed is introduced. And a seriesof attack behaviors are designed from simpleness to complexity. At last, the attackstrategy based on multi-robot cooperation is set up, which includes the attack mode ofthe main attacker and the situation of the assistant attacker.