Skip to main content
QUICK REVIEW

[论文解读] Towards Explainable Deep Learning for Ship Trajectory Prediction in Inland Waterways

Tom Legel, Dirk Söffker|arXiv (Cornell University)|Mar 4, 2026
Maritime Navigation and Safety被引用 0
一句话总结

该论文提出三种基于对内河船舶的交互感知的 LSTM 多船轨迹预测模型变体,这些变体将船舶领域权重与注意力机制分离,分析学习到的船舶领域值的可解释性,并在约 5 分钟内的与基线相比具有相近的 FDE 约 40 m。

ABSTRACT

Accurate predictions of ship trajectories in crowded environments are essential to ensure safety in inland waterways traffic. Recent advances in deep learning promise increased accuracy even for complex scenarios. While the challenge of ship-to-ship awareness is being addressed with growing success, the explainability of these models is often overlooked, potentially obscuring an inaccurate logic and undermining the confidence in their reliability. This study examines an LSTM-based vessel trajectory prediction model by incorporating trained ship domain parameters that provide insight into the attention-based fusion of the interacting vessels' hidden states. This approach has previously been explored in the field of maritime shipping, yet the variety and complexity of encounters in inland waterways allow for a more profound analysis of the model's interpretability. The prediction performance of the proposed model variants are evaluated using standard displacement error statistics. Additionally, the plausibility of the generated ship domain values is analyzed. With an final displacement error of around 40 meters in a 5-minute prediction horizon, the model performs comparably to similar studies. Though the ship-to-ship attention architecture enhances prediction accuracy, the weights assigned to vessels in encounters using the learnt ship domain values deviate from the expectation. The observed accuracy improvements are thus not entirely driven by a causal relationship between a predicted trajectory and the trajectories of nearby ships. This finding underscores the model's explanatory capabilities through its intrinsically interpretable design. Future work will focus on utilizing the architecture for counterfactual analysis and on the incorporation of more sophisticated attention mechanisms.

研究动机与目标

  • 为拥挤的内河航道中的准确轨迹预测提供动力,以提升安全性和交通管理。
  • 研究交互感知预测方法及其在内陆场景中的可解释性。
  • 开发将交互加权与注意力分离的模型,以实现可解释性和对照分析。
  • 评估提升的预测是否来自真正的交互感知,而非偶然相关性。

提出的方法

  • 将多船轨迹预测建模为一个具有 LSTM 编码器-解码器和全局 Luong 风格注意力的函数。
  • 引入可训练的船舶领域参数张量 S(Γ,Θ,Φ),基于离散化遇碰类型对隐藏状态融合进行加权。
  • 定义船对船的关系值 Γ、Θ、Φ,来自横向距离、相对方向和距离变化率。
  • 开发三种模型变体:EA-DA、E-DA 和 E-DDA,将交互加权与注意力分离,且 E-DDA 进一步把盲解码器与有注意力的解码器分离。
  • 在 Rhine AIS 数据(3 年,2021–2024)上按预测时间步长评估最终位移误差(FDE)。
  • 与一个与交互无关的基线进行比较,并分析船舶领域参数以提高可解释性。

实验结果

研究问题

  • RQ1可学习的船舶领域参数是否能有效指示哪些遇碰类型会影响目标船舶的未来轨迹?
  • RQ2将交互加权与注意力机制分离是否在不牺牲预测准确性的前提下提升可解释性?
  • RQ3在不同遇碰类型的内河场景中,不同模型变体(EA-DA、E-DA、E-DDA)表现如何?
  • RQ4预测改进是否来自实际的交互感知,还是由于模型的其他因素?
  • RQ5学习到的船舶领域值能揭示在内陆场景中对相对向后/同向船舶的注意力分配吗?

主要发现

  • 最终 5 步时间窗的 FDE 约为 40 米,所提模型在与相似研究的对比中具有竞争性。
  • E-DA 在各个时间窗上比 EA-DA 与无交互基线得到更低的误差。
  • E-DDA 优于 EA-DA,体现将目标处置与邻近处理分离在注意力中的好处。
  • 学习到的船舶领域参数并未始终与直观的交互重要性对齐,表明准确性提升并非仅来自交互感知。
  • 分析表明该架构具有可解释能力,能够进行潜在的对照分析,尽管注意力机制需要进一步优化。

更好的研究,从现在开始

从论文设计到论文写作,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。