Research of cooperative behavior in multi-robots system

The overlap of actions selected by each robot results in poor cooperation in multi-agent robotic system, at the same time the conventional reinforcement learning requires a large computation cost because each agent must learn. Using new method to solve these problems, the parametric control and enhancing technique will make the optimal control problem to be transformed into a series of parameter optimization problem, and the decision tree module can accelerate the learning process, so that the reinforcement learning module can choose quickly a proper cooperative behavior. The results of simulation indicate the effectiveness of the proposed method.