A nonlinear MPC based on application of a sequential distributed accelerated gradient method to the continuation method

The C/GRES method is known as an effective algorithm for numerical solution of optimal control problem in nonlinear model predictive control (MPC). When we apply the C/GMRES method, we usually assume that dynamics of the system model is expressed as equations analytically, however practical models sometimes contain maps and the dynamics of them is provided only numerically. In this paper, we propose a nonlinear MPC algorithm for such systems by applying a sequential distributed accelerated gradient method to the continuation method which is used in C/GMRES method.