A comparative study for balancing and positioning of an inverted pendulum robot using model-based controllers

Abstract

Abstract: This paper presents a comparison between four control approaches to solve the problem of balancing and positioning of the inverted pendulum robot (IPR). These control approaches are proportional-integral-derivative (PID), linear quadratic tracker (LQT), linear model predictive control (LMPC) and nonlinear model predictive control (NMPC). The first three approaches directly depend on the linear control theory and thus on the dynamic linearization around the stationary point. However, the last control approach depends on the original nonlinear dynamics, therefore, a superior control performance is obtained here. In this study, we test the IPR using these control approaches. In particular, these control approaches are simulated using Simulink/Matlab dedicated toolboxes like optimal control toolbox and MPC toolbox to test the linear control approach, in addition, a combined multiple shooting with collocation on finite elements method for nonlinear approach. Moreover, some of these scenarios are tested experimentally using a simple lab-made prototype.