Extended State Observer Based Reinforcement Learning and Disturbance Rejection for Uncertain Nonlinear Systems

This paper investigates the extended state observer (ESO) based reinforcement learning (RL) and disturbance rejection for uncertain nonlinear systems having non-simple nominal models. The ESO is first designed to estimate the system state and the total uncertainty. Based on the output of the observer, the control compensates for the total uncertainty in real time, and simultaneously, online approximates the optimal policy for the compensated system using a simulation of experience based RL technique. Rigorous theoretical analysis is given to show that the widely-used restrictive persistence of excitation (PE) condition is not required in the established framework. Simulation results are presented to illustrate the effectiveness of the proposed method.