Long-term Vessel Prediction Using AIS Data
Abstract
In recent years there has been much research dedicated to the development of autonomoussurface vehicles (ASVs), including large-scale autonomous ships. An importantpart of this research is to ensure that an ASV can safely operate in areaswith other marine traffic. Therefore it is necessary that an ASV is equipped witha collision avoidance (COLAV) system. A vital part of this system is the ability topredict the trajectories of other vessels in order to avoid them. This thesis is focusedon improving recently developed prediction techniques and investigate whether theyare suitable for use within a COLAV system.
In this thesis, the neighbor course distribution method (NCDM), which makes predictionsbased on automatic identification system (AIS) data and represents obstaclesas Gaussian mixture models (GMMs), is improved upon to make it more suitable foruse in COLAV. This method has shown promise, but as it is a data-driven methodit performs poorly when little data is available and is often overconfident. Therefore,a modified NCDM, which assumes near constant velocity in low data density areas,is developed in this thesis to mitigate these problems. The modified NCDM showsa significant improvement in covariance consistency, although there is some loss inaccuracy.
Furthermore, the modified NCDM is used for obstacle prediction within a COLAValgorithm based on model predictive control (MPC). This is benchmarked againstusing the same COLAV algorithm, but assuming that the vessels maintain a constantvelocity and modeling these obstacles as circular constraints. The new method isfound to be advantageous in areas where vessels historically have maneuvered often.The results also show that the modifications to the NCDM makes the algorithmsignificantly more suitable for COLAV than the original NCDM in areas with lowdata density.