Optimal Control of a Two-Wheeled Self-Balancing Mobile Robot Based on Adaptive Dynamic Programming

Abstract

This paper considers optimal tracking control for a two-wheeled self-balancing mobile robot with unknown dynamics. The aim is to achieve asymptotic tracking and disturbance rejection by minimizing some predefined performance index. Through the combination of adaptive dynamic programming (ADP) and internal model principle, an approximate optimal controller is iteratively learned online using measurable input/output data. Unmeasurable states are also reconstructed from input/output data. The discrete-time algebraic Riccati equation is iteratively solved by ADP approach. Simulation results demonstrate the feasibility and effectiveness of the proposed approach.