Maritime vessel trajectory prediction improved with informer model

TempTPI: Informer-Based trajectory prediction for maritime vessels

Machine Learning

Summary

Predicting where ships will go in the next few hours is important for safety and planning. The authors created a new computer method called TempTPI that uses a special Transformer model known as Informer to make these predictions more accurate and faster. They also added a way for the model to understand daily and seasonal travel patterns. Tests showed TempTPI predicts ship locations better than previous methods, especially over longer times like five hours ahead.

What this means in practice

  • For maritime traffic controllers: Improve long-term route planning and collision avoidance by accurately forecasting vessel paths up to five hours ahead using TempTPI.
  • For logistics and shipping planners: Enhance scheduling and resource allocation with better predictions of ship arrivals and movements based on TempTPI's model of vessel trajectories.

Tested on one dataset.

Authors

Kevin Ferneding, Veronika Lietavcova, Aleksandra M. Blachowiak, Peder Heiselberg

Abstract

Accurate long-term trajectory prediction for maritime vessels is essential for safety and logistical efficiency. While deep learning models, particularly Transformers, have shown promise in processing Automatic Identification System (AIS) data, they often struggle with the quadratic computational complexity of self-attention and the loss of accuracy over extended forecasting horizons. This study proposes TempTPI, a novel prediction framework that integrates an Informer-based encoder with a multi-channel temporal encoding mechanism. The Informer architecture leverages a ProbSparse self-attention mechanism to reduce computational overhead and focus on the most significant dependencies, while the temporal encoder utilizes Fourier-like frequency expansions to capture cyclic patterns (hourly, daily, and seasonal) in vessel behavior. We evaluate our model against the state-of-the-art TPTrans architecture using AIS data from Danish waters. Experimental results demonstrate that TempTPI consistently outperforms existing methods across prediction windows of 1 to 5 hours. Notably, at a 5-hour horizon, the proposed model achieves a 55% improvement in Mean Squared Error (MSE), offering a robust solution for long-range maritime situational awareness.