Dynamic Graph-Like Learning with Contrastive Clustering on Temporally-Factored Ship Motion Data for Imbalanced Sea State Estimation in Autonomous Vessel

TL;DR

The paper tackles sea state estimation for autonomous vessels, addressing severe data imbalance and time-series redundancy in ship-motion data. It introduces the Temporal-Graph Contrastive Clustering Sea State Estimator (TGC-SSE), a four-module framework combining Time Dimension Factorization, Dynamic Graph-like Learning, and a Contrastive Clustering Loss to jointly learn robust representations and balanced classifications. Empirical results across 14 UEA benchmarks and ship-motion datasets show state-of-the-art performance, with ablations confirming the positive impact of each module and the proposed loss. The method demonstrates strong generalization and practical potential for safer, more efficient autonomous marine operations. The work also highlights a clear trade-off between computational cost and accuracy, and suggests future extensions to regression and frequency-domain analysis.

Abstract

Accurate sea state estimation is crucial for the real-time control and future state prediction of autonomous vessels. However, traditional methods struggle with challenges such as data imbalance and feature redundancy in ship motion data, limiting their effectiveness. To address these challenges, we propose the Temporal-Graph Contrastive Clustering Sea State Estimator (TGC-SSE), a novel deep learning model that combines three key components: a time dimension factorization module to reduce data redundancy, a dynamic graph-like learning module to capture complex variable interactions, and a contrastive clustering loss function to effectively manage class imbalance. Our experiments demonstrate that TGC-SSE significantly outperforms existing methods across 14 public datasets, achieving the highest accuracy in 9 datasets, with a 20.79% improvement over EDI. Furthermore, in the field of sea state estimation, TGC-SSE surpasses five benchmark methods and seven deep learning models. Ablation studies confirm the effectiveness of each module, demonstrating their respective roles in enhancing overall model performance. Overall, TGC-SSE not only improves the accuracy of sea state estimation but also exhibits strong generalization capabilities, providing reliable support for autonomous vessel operations.

Figures (10)

• Figure 2: The pipeline of the proposed model.
• Figure 3: In the figure, the Contrastive clustering loss function is depicted. Each shape corresponds to a unique sea state class. During training, classes with a smaller sample size receive increased weighting. The thick arrow is denoted as ‘$L_{\text{cluster}}$’, the orange bi-directional arrow as ‘$L_{\text{neg}}$’, and the green line represents ‘$L_{\text{pos}}$’.
• Figure 4: Sensitivity Analysis on AtrialFibrillation Dataset.
• Figure 5: Ablation study.
• Figure 6: Sensitivity analysis