As maritime shipping grows, ensuring navigation safety becomes increasingly vital. Accurate trajectory prediction clarifies a ship’s future intentions, supporting safe navigation. In recent years, many researchers have focused on extracting spatial and temporal features to improve trajectory prediction accuracy. Although these methods have shown some predictive performance, they often overlook frequency domain features, significantly limiting their predictive potential. To this end, Discrete Wavelet Transform (DWT) is applied to ship-trajectory data for time–frequency conversion, capturing both global trends and local dynamics. Furthermore, a Frequency-Domain Interaction Space (FDIS) is constructed to aggregate and exchange features across different frequencies, thereby improving information utilization. Finally, we evaluate the proposed model against the most advanced prediction methods on three real-world AIS datasets from multiple perspectives. Experimental results demonstrate that the proposed model consistently achieves the highest prediction accuracy across various scenario.
Song et al. (Mon,) studied this question.