Reinforcement Learning with Perturbation Method to Turn Unidirectional Linear Response Fuzzy Controller for Inverted Pendulum

摘要

In this paper, we present a unidirectional linear response fuzzy controller (FC) to control the inverted pendulum system. The performance of turning fuzzy controller is defined as an evaluation function and our proposed technique, which is based on the integration of reinforcement learning and a perturbation method, is utilized to diversity the search of minimization of the evaluation function. Once the reinforcement learning method results in a local optimal solution for the problem, the perturbation method is implemented, actually a gradient ascent learning phrase is carried out. The two phrases are repeated until a better solution is found. The proposed hybrid learning method is utilized to adapt the parameters in the fuzzy control. Simulation results reveal that the proposed learning system can control the inverted pendulum system better than other traditional systems.