Model Predictive Tracking Control for USV with Model Error Learning

Abstract

AbstractThis paper is concerned with the learning-based model predictive control (MPC) for the trajectory tracking of unmanned surface vehicle (USV). The accuracy of system model has a significant influence on the control performance of MPC. However, the complex hydrodynamics and the complicated structure of USV make it difficult to capture the accurate system model. Therefore, we present a learning approach to model the residual dynamics of USV by using Gaussian process regression. The learned model is employed to compensate the nominal model for MPC. Simulation studies are carried out to verify the effectiveness of the proposed method.KeywordsModel predictive controlTrajectory trackingGaussian process regressionUnmanned surface vehicles