ADP–based control of a two–wheeled self–balancing mobile robot

Abstract

This study investigates optimal control for a two-wheeled, self-balancing mobile robot whose dynamics is unknown. The objective is to achieve asymptotic control and rejection of disturbances while minimizing a certain index of performances. An approximate optimum controller may be iteratively learnt online utilizing quantifiable input and output data by combining the internal model concept with adaptive dynamic programming (ADP). States that cannot be measured directly are additionally recreated using this information. The ADP method is used to solve the discrete-time algebraic Riccati problem iteratively. The efficacy and viability of the suggested approach are shown by the simulation results.