Abstract
In maritime traffic management, the precise prediction of vessel trajectories is paramount, given the industry’s substantial dependence on vessel transportation for the transport of commodities, passengers, and energy resources. This study proposes two innovative prediction methodologies (short-term and long-term) for vessel movements. Furthermore, we introduce a novel evaluation metric designed to quantitatively assess the efficacy of the proposed short-term prediction method in forecasting vessel trajectories. The presented methodologies were empirically tested, employing two-month Automatic Identification System (AIS) data collected from the Baltic Sea to examine their performance. Preliminary experimental outcomes indicate a superior level of accuracy embodied in the short-term prediction method. On the other hand, the long-term prediction method demonstrated enhanced performance metrics in the context of computational speed and memory utilization. These observations underscore the potential of the proposed methodologies to amplify efficiency and augment safety standards in marine traffic management.
Javad Sheikh
Project Researcher at
I’m an engineer and programmer passionate about solving problems with machine learning and deep learning.