The maritime industry, responsible for over 90% of global trade, faces growing challenges in safety and efficiency due to increasing traffic volume and complex navigational environments. Vessel Trajectory Prediction (VTP) has thus become a critical technology for enabling intelligent maritime management. This paper provides a comprehensive review of the field, systematically analyzing the evolution of prediction methodologies from traditional approaches to advanced deep learning and hybrid models. While the proliferation of Automatic Identification System (AIS) data has enabled data-driven trajectory analysis, significant challenges persist, including data quality issues, model capability limitations in capturing complex spatio-temporal dependencies, and generalizability across diverse maritime scenarios. This review not only synthesizes existing achievements but also critically examines current limitations in generalizability, interpretability, and real-time capability. • This analysis addresses vessel trajectory prediction challenges, examining data quality issues, model limitations in spatiotemporal dependencies, and generalizability across maritime environments. • It reviews methodological evolution from physical models to deep learning and hybrid approaches, comparing their strengths. • The study integrates multi-source data frameworks, combining AIS with environmental factors like meteorology and behavioral patterns to boost accuracy. • Emerging directions include explainable AI, real-time prediction, and cross-domain generalization, paving the way for next-generation intelligent maritime systems.
Jiang et al. (Sun,) studied this question.