Embedded linear model predictive control through mesh adaptive direct search algorithm

In this paper we apply the derivative-free mesh adaptive direct search (MADS) algorithm to find the minimum of a constrained optimization problem, resulting from model predictive control (MPC). MPC requires indeed to solve an optimization problem online, at each sampling time of the system to regulate. A progressive barrier approach is used in MADS, in order to cope with the possibly infeasible initial point for the algorithm. Hardware-in-the-loop simulations are performed where the MADS-based MPC regulator is implemented on a microcontroller and a double integrator system is simulated on a PC. Control performances and circuit latency are assessed with respect to the number of MADS iterations