Fuzzy Reinforcement Learning Based Trajectory-tracking Control of an Autonomous Mobile Robot

The aim of this study is to achieve trajectory-tracking of an Autonomous Mobile Robot (AMR). The proposed Fuzzy Reinforcement Learning Control (FRLC) takes fuzzified sliding surfaces consisting of tracking errors and local velocities as inputs and generates actuator voltages as outputs. The relationship between inputs and outputs is reasoned by using the trained knowledge of a reinforcement learning agent. Simulations are conducted to investigate the performance of the proposed FRLC. It is shown that the proposed FRLC has excellent control performance for trajectory-tracking without reasoning human designed.