Constrained predictive control of a robotic manipulator using quasi-LPV representations

In this paper a practical approach to Nonlinear Model Predictive Control (NMPC) of a robotic manipulator subject to nonlinear state constraints is presented, which leads to a successful experimental implementation of the control algorithm. The use of quasi-LPV modelling is at the core of this scheme as complex nonlinear optimization is replaced by efficient Quadratic Programming (QP) exploiting the quasi-linearity of the resulting model and constraints. The quasi-LPV model is obtained via velocity-based linearization which results in an exact representation of the nonlinear dynamics and enables stability guarantees with offset-free control. The experimental results show the efficiency and efficacy of the algorithm, as well as its robustness to unmodelled dynamics