Supervisory Control System for Cooperative Agents

Abstract

In this paper it's proposed a new hybrid supervisory system (SS) for cooperative agent system's control. This supervisory system contains fuzzy logic on its many parts, as well as automatons theory, pattern recognition based on integral images, shape recognition and RGB matrix, and a non-linear fuzzy controller properly said. It's presented as an example the robot soccer competition, which each agent's action depends as much in another agent's reaction as on the cooperation of the others. Although the system is highly dynamic, the strategies that rule this system can be seen on a discreet view as a way to select the best strategy to be used easing the work of the motor's fuzzy controller. Also is proposed a control based on the change of the linear and angular velocities so the computational work can be drag down. The intelligent logic's implementation based on fuzzy logic is used in many ways such as zone characterization and robot's control