Research of Multi-Agent System Model and Learning Method

Learning ability and system model are key characteristics of agents, especially in multi-agent environment. Define logical grammar, semantic rule and important properties of agents, and establish a multi-agent robot system model M based on the criterion L. Propose a hybrid intelligent multi-agent cooperation learning algorithm to overcome the limitation of individual and group study. Present experiment conclusions on the basis of the model of special assembly robot. Experiment results show that the applications of the model and learning algorithm to the robot assembly system have obviously improved individual technology and team collaboration, and the distributed control mode can enhance reliability of the system.