Autopilot with adaptive vessel modelling

A feasibility study of an application of adaptive model in a ship autopilot is presented. The system for predicting a ship motion, developed in the previous studies, was enriched by a control algorithm of the ship motion and the adaptive model of the ship. The signals from the navigation sensors (DGPS, log, and gyrocompass) are filtered by dedicated Kalman filters, and then processed by an integration algorithm to obtain the best estimation of the actual ship position, heading, velocity and rate. These data form the state vector of the ship dynamic model. The model of the ship dynamics has parameters which are identified on-line using particle filter method. The efficiency of applied control methods and collaboration with on-line adaptive dynamics modeling was verified by simulations.