NMPC Combining Parameter Identification on Autonomous Marine Surface Vehicle

In recent times, Nonlinear Model Predictive Control (NMPC) has risen in prominence as a favored technique for managing the complex nonlinear dynamics of Autonomous Marine Surface Vehicles (ASV). However, there are still challenges with NMPC when ASVs are operated under unknown environmental disturbances like ocean currents. In general, there are two methods to address this type of problem: one is to design a robust NMPC controller in spite of environmental uncertainties while another method is to carefully identify environmental uncertainties beforehand. Unfortunately, both methods are not very practical. A robust NMPC controller may end up a difficulty of establishing recursive feasibility and it is not realistic to identify environmental uncertainties beforehand either. In this study, we propose and implement a new framework to control ASV using a rigorous NMPC controller concurrently combining parameter identification with maximum likelihood estimation (MLE) in order to identify and update environmental uncertainties. In addition, there is a strong demand for a quick computer program to implement both the NMPC and MLE when a fast response is required for controlling an ASV. Much of the computational demand in implementing NMPC and MLE stems from the evaluation of the gradient vector and the Hessian/Fisher matrix. This study employs sophisticated symbolic computation techniques to analytically determine these intricate components. The integrated NMPC with the MLE parameter estimation approach effectively addresses a two-dimensional ASV trajectory tracking challenge amidst uncertain ocean currents.