Research And Design Of MiroSot Soccer Robot Decision-making Subsystem

Robot soccer game is an international competition of advanced technology rapidly developed in these years. It is acombination between the athletics and the high-techs, whereinvolves many high and new-technology industries such as robotics, electric-mechanical integrated integration technology, communication and computer science, the robot vision and sensor fusion technique, the decision-making and countermeasure -making concept, and intelligent control. As a technique 'war' on a confined platform, the robot soccer is interested by the more and more researchers.MiroSot system is such a soccer robot system based on the vision, which consists of the vision subsystem, the decision -making subsystem, the communication subsystem and the robot-body subsystem. Herein the decision-making subsystem iscollaboration of the robot players and their path-planning during the whole competition. Whether or not the decision-making subsystem is designed well, it is the linchpin upon which the success or failure in the robot soccer competition depends. This thesis deals with the research and design of the decision-making subsystem based on the robot soccer platform from the following aspects.1) The Formation Selection and the Role Assignment Based on the analysis of the competition arena the formation selection and the role assignment are the key factors of guaranteeing to obtain an effective upper strategy in the decision-making subsystem. The good formation is favorable to the organization and flexibility of the competition. Taken the 3-to-3 match as an example, the rules of the formation are designed based on the arena conditions. Under the high dynamic environment, the fixed position of the robot player will give rise to the inflexibility and inorganization. The dynamic role assignment will enhance the operational competence of the robot team. Therefore, a design of the role assignment based on the fuzzy logic is brought forward for the purpose of the simplicity and the timeliness. The distance from the robot to the target point, the relative orientation between the robot and the target point and obstacles existing in them are taken as the three input variables to the fuzzy logic system. The output variable is chosen as whetheror not to obtain the role assignment and then the rules are decided to complete the effective fuzzy logic system.2) The Design of the Action Base of the Decision-Making SubsystemWith the formation and the role assignment confirmed, every soccer robot will response to the corresponding technical action. The action is the ultimate representation of the decision-making subsystem, which is the basis and guarantee to the decision -making. Whether or not the behaviors of the soccer-robot complete well, it concerns the holistic performance of the whole robot soccer system. According to the experience from the human football the action of the soccer-robot is divided into the basic behavior, the basic action and the tactical action, moreover all the kinds of actions are illustrated in the thesis.3) The Motion-Planning and Obstacle Avoidance in Robot SoccerThere are two hot spots and nodus in the soccer robotsystem, which are the motion-planning in the complex anddynamic environment and the obstacle-avoidance in the existence of the obstacles. It is concerned how to implement the strategies of motion-planning with satisfactory and how to arrive to the target points by avoiding the obstacles in the highly dynamic robot soccer competition. By summarizing the motion-planning approaches of the mobile robot and concerning the characteristicsof the soccer robot, the combination of the real-time ability and the shortest path will be a right solution which breaks through the bottle-neck problems existing in the motion-planning and dynamic obstacle avoidance. The artificial potential field (APF) approach has been usually used to implement the motion-planning of the robot soccer system. However the system is brought into the local extreme value and faces dead-locking problem by this method. The Genetic algorithm (GA) approach can also solve the motion-planning problem and get the better solution at the cost of time, hence is not often applied into the soccer robot system in practice. In the thesis, the author puts forward a novel algorithm for the motion-planning and obstacle avoidance based on enhanced APF and GA by optimally grouping these two approaches. The simulation of the algorithm is implemented inMatlab6.5(?) and the satisfactory result shows that the algorithm is effective.