Adaptive control of self-balancing two-wheeled robot system based on online model estimation

Abstract

In this article, an adaptive model predictive controller (MPC) is designed for the position control of the self-balancing two-wheeled robot system. The system future output is optimized using the MPC controller by computing the manipulated variable trajectory. Traditional MPC uses a Linear-Time-Invariant (LTI) dynamic model of the system for the prediction of future behavior. The model of the self-balancing two-wheeled robot system is strongly nonlinear which degrades the prediction accuracy of the traditional MPC controller. Therefore, an adaptive MPC controller is designed based on linear-time-varying Kalman filter which online tunes and updates the estimated system parameters and accordingly produces the control effort in the presence of the input/output and state constraints. The performance of the proposed controller is compared with the traditional MPC controller and PID controller. The results show improved reference tracking and better stability for the proposed adaptive MPC controller as compared to traditional MPC and PID controller.