Design of Model Predictive Control of two-wheeled inverted pendulum robot

Abstract

Model Predictive Control (MPC) is a control strategy that uses an explicit model of a process to optimize the performance of the system. This paper describes the design of MPC controller for a Two Wheeled Inverted Pendulum (TWIP) considered as an underactuated and nonlinear system. By using a decoupling unit, TWIP can be controlled independently by two MPC controllers. Due to the feed forward control of MPC, the performance of the control system is improved to compensate for measurable disturbances. The validity of the design is evaluated in the presence of step-like disturbance and two different external forces. Finally, the result of the MPC controller is compared with LQR controller.