Research of Reinforcement Learning Control of Intelligent Robot Based on Fuzzy-CMAC Network

Intelligent grasping by servo robot-hand of monocular vision is a complex nonlinear control problem. In order to achieve high accuracy, model structure and parameters design of Fuzzy-CMAC are studied in this paper, and improved reinforcement learning algorithm is introduced into FCMAC. According to the reinforcement signals, different strategies are used to improve convergence speed and algorithm optimization. Based on the theory research, we do experimental research and try FCMAC to approach the nonlinear part of the dynamic model of the robot-hand. By analyzing the experimental results, it is concluded that this control strategy can solve the nonlinear problem effectively and the trained robot-hand can grasp the target object fleetly and accurately.