CmIVTP：基于跨模态交互的航海智能航船轨迹预测
摘要
航海智能交通系统（MITS）是确保繁忙水域导航安全和效率的关键因素。然而，由于单源数据存在局限性，准确的航船轨迹预测仍然具有挑战性。自动识别系统（AIS）数据常常稀疏或无法获取小型航船数据，而闭路电视（CCTV）数据单独使用无法完全捕捉航船动态行为。为缓解这些挑战，我们提出了一种基于跨模态交互的航船轨迹预测框架（ CmIVTP），用于建模航船动力学与环境约束之间的复杂交互。具体而言，我们引入目标感知场景编码器，以提取场景语义特征，有效捕捉航船-环境交互，提高轨迹预测精度。此外，我们提出了跨模态交互变换器，将AIS派生的运动特征、CCTV基于环境的特征以及场景表征相结合。它利用跨模态注意力机制同时捕捉内部模态语义和跨模态交互，确保预测具有动态一致性和环境可行性。 Furthermore, we construct a vessel group trajectory bank by clustering historical AIS trajectories into representative motion patterns, providing an efficient and scalable approach for candidate trajectory generation. Additionally, we introduce the maritime multimodal dataset plus (named Maritime-MmD), a large-scale dataset that synchronizes AIS data and CCTV video data, providing robust support for multimodal trajectory prediction research. Extensive experiments demonstrate that CmIVTP achieves better performance on multimodal-driven vessel trajectory prediction benchmarks. The code resources for this work can be available at https://github.com/LouisYxLu/CmIVTP.
引用
@article{arxiv.2605.26524,
title = {CmIVTP: Cross-modal Interaction-based Vessel Trajectory Prediction for Maritime Intelligence},
author = {Yuxu Lu and Dong Yang and Xiaoyu Li and Mengwei Bao and Congcong Zhao},
journal= {arXiv preprint arXiv:2605.26524},
year = {2026}
}