Nonlinear model predictive control based on the best-step Newton algorithm

Newton based path planning has been successfully applied to driftless nonlinear systems. It has been extended to feedback implementation in a form similar to nonlinear model predictive control. The closed loop system is shown to be asymptotically stable under suitable conditions which are reasonable for driftless systems but restrictive for systems with drift. This work presents a modification of the algorithm that improves its applicability to systems with drift. The key idea is to update the predicted control sequence to reduce the M+1-step-ahead error instead of the M-step ahead error, where M is the length of the prediction horizon. A numerical examples is included to illustrate the efficacy of the proposed approach.