Developing a Cooperative Behavior for Multi Agents System Application to Robot Soccer

Abstract

Middle League MiroSot robot soccer system is to be used as a test bed for a multi-agent cooperative system to develop new cooperative strategies. The robot soccer game is different from other multi-agent systems, in that the robots in one team have to cooperate in the face of competition from the opponent team in a dynamically changing environment. The multi-agent control algorithm must comprise of low level kinematics and dynamics, high level strategies to avoid obstacles, cooperate among teammates, and to compete with the opponent robots. In such a dynamic environment, a robot needs fast processing algorithms and time-optimized team cooperative strategies. The current research is geared towards developing advanced innovative path generation and planning, enhanced game strategies, optimized moving ball intercept, and passing behavior. This should result in full functionality of American University of Sharjah (AUS) existing robot soccer system enabling AUS team to participate in the next FIRA Middle League Championship. To achieve the objectives of this study, one needs to comprehend the existing system and obtain full functionality of the hardware and the software components of the system. This requires that the existing strategy code must be analyzed to identify current used algorithm. In the second part the proposed strategy must be created, applied and compared with current strategy. Performance measures should be obtained to evaluate the performance gains. These performance measures should include speed of the game, precision in scoring goals, and robustness of the game.