Fuzzy System Centered Obstacles Avoidance Path Planing In MIROSOT Soccer Robot System

Establishing the new standards for the artificial intelligence(AI) and the robotics, bridging basic research and practical application, bringing forth harvest of information automatic development and advancing new methods of produce, study and research, soccer robot has the nice future about the huge industrialization.MIROSOT robot soccer system based on vision is the remote-brainless and multi-robot cooperating soccer robot system. The goal of the multi-robot path planning is to make out, for every robot, the routes which are collision- free with the obstacles and other robots. Robotic obstacle avoidance is the main part of the path planning of the robotsThe main research content in this paper is the robotic action of obstacles avoidance, on the MIROSOT soccer robot system, which adopts the distributed planning method. That is to say the method is the way of self-planning the trace for every robot and the planning idea is behavior centered.Obstacles avoidance based on potential field is general method the route planning adopts. But this method neglects a problem that the obstacles from different direction and distance do not take thesame influence to the avoidance trace of robots, and it takes the same way to deal with those obstacles so that it results in huge calculations, slow response and twisted traces.The fuzzy system is the knowledge-centered or rule centered system. In the soccer robot system, there is much information in the different degree which is not certainty, forecasted and exact measured. Considering of the obstacles avoidance of the robot, this paper pays attention to two factors:(1) Danger degree between the obstacles and the robots(2) Distance degree between the goal and the robotsRobots should move along the trace whose danger degree and distance degree are both slower, and try to select smooth trace.The number of fuzzy sets about fuzzy variable and fuzzy rules takes the influences to the approximation accuracy of fuzzy sets. In order to escape the bondage from curse of dimensionality, this paper adopts Gaussian fuzzifiers and intersection inference engine. This method not only takes the non-linear processing, but results in simple calculation and rapid response of the system.Fuzzy inference centered robot moves along the direct which is multi-angle with the direction between the goal and robots, and the avoidance trace is not smooth or continuous. In order to get a smooth and continuous trace, this paper uses the least squares algorithm to imitate the curve which is reasoned from the fuzzy system and consisted with the distributed point set. So this curve adapts to the motion and control of the robot to avoid obstacles.