Suboptimal model predictive control of a laboratory crane

A fast model predictive control (MPC) scheme is presented and applied to a laboratory crane with five degrees–of–freedom. The MPC scheme accounts for control constraints and is based on the gradient projection method that allows for a time and memory efficient computation of the single iterations. To guarantee real-time feasibility, a fixed number of iterations is used per sampling step. Although this leads to a suboptimal solution, its application to the nonlinear crane model reveals the performance as well as the high computational speed of the method. The feasibility of the proposed approach is shown by means of experiments of a laboratory crane with a sampling time of 2 ms on a standard real-time hardware.