Comparison of stabilizing NMPC designs for wheeled mobile robots: An experimental study

Abstract

In this paper, two stabilizing nonlinear model predictive control (NMPC) designs, namely, final-state equality constraint stabilizing design and final-state inequality constraint stabilizing design have been applied to achieve two wheeled mobile robot's control objectives, i.e. point stabilization and trajectory tracking. In both controllers, final-state constraints are imposed, on the online optimization step, to guarantee the closed loop stability. As shown in the literature, both stabilizing designs were addressed to be computationally intense; thus, their real-time implementation is not tractable. Nonetheless, in this work, a recently developed toolkit implementing fast NMPC routines has been used to apply the two stabilizing designs on a mobile robot research platform after developing a C++ code, coupling the toolkit and the research platform's software. Full scale experiments implementing the two stabilizing designs are conducted and contrasted in terms of performance measures and real-time requirements.