LLM4STP: A large language model-driven multi-feature fusion method for ship trajectory prediction

Jiao, H ORCID: 0000-0002-2108-0885, Gong, J ORCID: 0009-0007-6450-7109, Li, H ORCID: 0000-0002-4293-4763, Lam, JSL ORCID: 0000-0001-7920-2665, Shu, Y ORCID: 0009-0001-2529-1516, Wang, J ORCID: 0000-0003-4646-9106 and Yang, Z ORCID: 0000-0003-1385-493X (2025) LLM4STP: A large language model-driven multi-feature fusion method for ship trajectory prediction. Transportation Research Part E: Logistics and Transportation Review, 207. ISSN 1366-5545

Preview

Text LLM4STP- A large language model-driven multi-feature fusion method for ship trajectory prediction.pdf - Published Version Available under License Creative Commons Attribution. Download (10MB) | Preview

Abstract

Ship trajectory prediction (STP) is a critical research focus for enhancing maritime traffic situational awareness and supporting navigational decision-making in intelligent transportation systems. The accuracy and robustness of prediction models significantly affect maritime safety and shipping efficiency. Despite advances driven by Automatic Identification System (AIS) data and deep learning techniques, key challenges remain unresolved, including dynamic multi-ship interaction modelling in complex marine environments, multi-scale temporal dependency reasoning, trajectory uncertainty quantification, and effective integration of maritime domain knowledge. Existing methods based on Large Language Models (LLMs) improve generalisation through pre-trained knowledge but fall short in real-time interaction topology modelling, geospatial semantic representation, and uncertainty estimation. To address these limitations, this paper proposes LLM4STP, a novel LLM-driven multi-feature fusion method for STP. LLM4STP establishes a new paradigm by deeply integrating LLMs with maritime domain knowledge to collaboratively predict ship trajectories. The model features an adaptive graph-masked Transformer to dynamically capture ship interaction topologies, hierarchical temporal reasoning to jointly model local manoeuvring behaviours and macroscopic navigational intent, and an innovative fusion of Gaussian probability distribution heatmaps with GeoHash-based geospatial encoding to quantify trajectory uncertainty while preserving semantic continuity. Experiments on three representative water areas demonstrate that LLM4STP consistently outperforms nine state-of-the-art (SOTA) methods, as validated by key metrics including Average Displacement Error (ADE), Fréchet Distance (FD), and Final Displacement Error (FDE). Moreover, the few-shot learning experiments demonstrate that LLM4STP can match the performance of models trained on the full dataset using only 20 % of the training data, highlighting its efficiency and strong adaptability in data-scarce environments. The ablation studies empirically validate the significance and distinct contribution of each component within the proposed model architecture. This study integrates LLM into maritime traffic scenarios, making significant contributions to enhancing the robustness, accuracy, and interpretability of STP in high-interference environments. The source code is openly accessible at https://github.com/Joker-hang/LLM4STP .

Item Type:	Article
Uncontrolled Keywords:	Large language model; Ship trajectory prediction; Situational awareness; Hierarchical temporal modeling; Trajectory uncertainty; GeoHash encoding; 3509 Transportation, Logistics and Supply Chains; 35 Commerce, Management, Tourism and Services; Machine Learning and Artificial Intelligence; 14 Life Below Water; 7 Affordable and Clean Energy; 0102 Applied Mathematics; 0103 Numerical and Computational Mathematics; 1507 Transportation and Freight Services; Logistics & Transportation; 3509 Transportation, logistics and supply chains
Subjects:	Q Science > QA Mathematics > QA76 Computer software T Technology > TA Engineering (General). Civil engineering (General)
Divisions:	Engineering
Publisher:	Elsevier
Date of acceptance:	5 December 2025
Date of first compliant Open Access:	27 April 2026
Date Deposited:	27 Apr 2026 12:15
Last Modified:	27 Apr 2026 12:15
DOI or ID number:	10.1016/j.tre.2025.104599
URI:	https://researchonline.ljmu.ac.uk/id/eprint/28468

View Item