Target Identification And Obstacle Avoidance Planning Of Mirosot Soccer Robot

As the twentieth century's new competition, soccer robot attracted worldwide attention. Micro Robot World Cup Soccer Tournament (MiroSot) system involves the study of robot, pattern recognition, image processing, digital signal processing, electromechanical integration, and other disciplines, therefore it became an important experimental platform for multi-agent system and a hot spot of research domestic and abroad.In this description, based on MiroSot 3VS3 soccer robot system platform, the visual system target identification and obstacle avoidance planning are studied and simulated.First, the structure of robot soccer system and working principle of vision subsystem are introduced. In robot soccer tournament, the primary task of robot is to find the other robot and the target ball, only in this way, obstacle avoidance and other actions can be done. Based on rapid HSI transform space, a target identification algorithm is adopted, which combined a mobile grid and dynamic window technology. Based on progressive PHT, the other fast target identification algorithm was proposed,which extract ROI(Region of Interest) firstly, run PPHT secondly. Computation was reduced ,the antinoise and real-time performance were improved.Second, according to the actual requirements of robot soccer competition, the obstacle avoidance planning is studied. The velocity vector is brought to the potential field. Then a kind of new repulsive potential function and attractive potential function are deduced, under the effect of the new potential functions, the robot can avoid obstacles and move to target quickly.Finally, according to the demand of experiment and software development, the MiroSot robot system software was analyzed and studied. The source code of the system was written by MFC of Visual C++ 6.0. The software structure of vision subsystem was mainly analyzed. Each function modules are explained and debugging.