Machine Learning-Based Estimation Of Wave Direction For Unmanned Surface Vehicles

TL;DR

The paper tackles estimating wave direction for USVs by framing it as a regression task solved with an LSTM-based model that ingests sequential sensor data and outputs the sine and cosine components of the wave direction. It introduces a three-layer architecture (data processing, LSTM core, and dense output) and a thorough data-processing pipeline that handles transect segmentation and angular discontinuities. Across pool and open-sea experiments, the method outperforms baseline sequence models in both MAPE and angular accuracy, demonstrating robust generalization to unseen data. The work highlights practical implications for real-time USV navigation safety and proposes future work on richer field data, real-time deployment, and reinforcement learning for autonomous behavior in dynamic maritime environments.

Abstract

Unmanned Surface Vehicles (USVs) have become critical tools for marine exploration, environmental monitoring, and autonomous navigation. Accurate estimation of wave direction is essential for improving USV navigation and ensuring operational safety, but traditional methods often suffer from high costs and limited spatial resolution. This paper proposes a machine learning-based approach leveraging LSTM (Long Short-Term Memory) networks to predict wave direction using sensor data collected from USVs. Experimental results show the capability of the LSTM model to learn temporal dependencies and provide accurate predictions, outperforming simpler baselines.

Figures (7)

• Figure 1: Model overview - A temporal view
• Figure 2: Experimental settings for data collection
• Figure 3: Drone's trajectory in experimental pool
• Figure 4: Drone's trajectory in open sea
• Figure 5: Sequence size effect on model performance