The Research Of Strategy System Of Soccer Robot

With the development of robotics, robots are used more and more widely. Since the 80's, multi-robot system enter people's life with higher precision, load-bearing capability and flexibility than a single robot, and more and more people were concerned with it. Multi-robot group's problem of organizations and control has been put forward with the development of robot system application. Multi-robot cooperation is an important part of the system which will give full play to the advantages of multi-robot system.This thesis focuses on research of the AI strategy system of RoboCup small-size soccer robot system. Strategy system uses multi-layer structure. According to the characteristics of multi-layer structure and functions, the system is divided into layers of information pro-processing layer, coordination layer, task layer and action layer. The level of the system is clear and easy to be implemented with this multi-layer structure. The system establishes Playbook by a number of different Plays based on multi-agent coordination mechanism (such as the Offensive Play, Defensive Play, Discipline Play and Waiting Play). The field is divided into several regions by static region division method. Because there are not a lot of region, play choice and coding work are not very heavy, and plays would not switch so frequently from region to region as to affect the speed of aggression. It can prevent robot targets from hopping effectively by adding flexible regions. Using combination of the distance and angle between robots and the ball, we proposed two-point cooperation mechanism which can bring multi-robot cooperation into effect.In order to solve the robot control latency problem of real time system, we found a method to measure system delay, using linear prediction and BP neural network method to establish prediction model of robot and ball. By analysis we found that the accuracy has been increased.We established the dynamic and kinematic model of robot, developed the motion control algorithm based on the theory of optimal control. Both simulation and experimental result show that it can track arbitrary path quickly and effectively.This thesis takes the RoboCup small-size robot soccer as a platform, has carried on the simulation and experiment to test strategy system. The experiment result indicated that the performance of the whole strategy system has been improved, mainly is: as the prediction precision is improved, the control of robots is more accurate and the capability of goalkeeper and robot interception is also enhance. The multi-robot strategy increased multi-robot cooperation, which improved the troop cooperation attack ability and made robots complete the task better.