Study On Centralized-control Robotics Soccer Decision-making System

As a challenging and standard subject in the field of artificial intelligence, the soccer robotgame has received the world-wide recognition and interests. Besides, the development of thestudy in soccer robot is a strategic step from basic theories to practical application. What’s more,as heart of the whole robotics soccer system, the strategy subsystem has interfaces with bothvision subsystem and communication subsystem, whose inferior and superior influence thereal-time and the accuracy of the robots’ movement and also play a decisive role in improvingthe whole game performance. Therefore, researching and designing strategic and tactical of thestrategy system as well as improving and optimizing the key algorithms, which can make itbetter to adapt to the complicated, dynamic, high real-time and difficult to get the accurate modelgame environment, have the theoretical significance as well as application value.In this paper, decision-making subsystem’s research work was launched based on RoboCup5VS5Small Scale Robot Soccer platform which was autonomously developed by ZhejiangUniversity. In this study, a thought of three-tier structure model for decision-making subsystemof robot soccer was put forward and then there was a detailed exposition and analysis for eachlayer; Library of strategy knowledge was also studied and designed, raising several importantdecision-making patterns and several kinds of robot soccer’s intelligent action implementationmethods. In view of the system delay’s influence on game performance caused by theclosed-loop control system of robot soccer, an improved extended Kalman filter algorithm wasproposed to predict the ball position and the motion model for the ball position prediction wasestablished; and then the superiority of the algorithm was proved through MATLAB simulationand physical experiments. The chapters in the thesis are summarized as follows:The first two chapters focused primarily on the following contents: Firstly, there was a briefdescription for the generation of the robot soccer, development status at home and abroad, pros-pects and its significance, summarizing and analyzing the research status and development trendsof the decision-making subsystem. In chapter II, after the analysis of the working principle of the soccer robot, there were an outlined description of the system components and working processfor the RoboCup small scale robot soccer and also a detailed instruction for the functions ofvarious subsystems of the robot soccer as well as the relationship between them. At the sametime, the problems and key technologies which need to address in each subsystem were also laidon.The third chapter is the key of whole thesis, on the basis of strategy general design, athree-tier decision-making structure model was proposed, focusing on the strategy knowledgelibrary designing and researching. The fourth chapter put forward a kind of extended Kalmanfilter algorithm to predict the ball position in the robot soccer system. On the basis of the detailedresearch on Kalman filter algorithm, an improved method was set out, establishing the motionmodel for the ball position prediction.The fifth chapter is the experiments and results analysis section. Simulations werecarried out on simulation platform for the two local strategy proposed in chapter three, provingthe feasibility and effectiveness of these two tactics; There were also Matlab simulationexperiments for algorithms proposed in chapter fourth; In addition, the extended Kalman filteralgorithm was also applied in the individual robot, such as kicking and shooting, to prove thesuperiority of this new algorithm. It was proved that the algorithm can use in the small robot soc-cer’s real time game environment and played a vital role in improving the performance of thegame.The last part is the conclusion and the prospect of this subject, there is a summary for theachievements of this study, pointing out the further improvements and the research outcomes atthe same time.