Vessel Route Prediction from AIS Data

Abstract

Route prediction for vessels when implemented on a vessel traffic monitoring solution can be used to prevent collisions and illegal activities from happening. Since route prediction will provide in-depth knowledge on a ship’s route at least for a few minutes, monitoring officers and ship captains will know an approximated location of all ships after a few minutes. This will help them to find out if there is a possibility of ships will collide with each other or with an object on ground before few minutes. This will help them to change ship’s course and speed in order to avoid collisions. Vessel Traffic Management System (VTMS) which was developed locally was the subject of this research. Even though the system already has a path prediction system, it is not accurate enough to carry out predictions when the ships move at higher speeds and changing their courses quickly. Hence there is a need of better prediction model for this VTMS. Even though most of the commercial vessel traffic monitoring systems have path prediction algorithms with them, they are not available for public. Thus, this research explores an efficient method of predicting vessel paths using Kalman Filter based techniques. These techniques use previous data when predicting the path. Hence the database of VTMS was used as the test data for this research. Kalman filter and its variation are very popular methods for the solutions to prediction related problems. Thus, Kalman filter based solution was developed and tested with test cases. Developed solution was able to tackle the objective of this research giving better performances and accuracy.