PTrajM: Efficient and Semantic-rich Trajectory Learning with Pretrained Trajectory-Mamba
Yan Lin, Yichen Liu, Zeyu Zhou, Haomin Wen, Erwen Zheng, Shengnan Guo, Youfang Lin, Huaiyu Wan
TL;DR
PTrajM introduces Trajectory-Mamba, a movement-behavior–parameterized encoder built on Traj-SSM to efficiently model continuous motion from irregular trajectory points, and a travel purpose–aware pre-training that aligns trajectory embeddings with road and POI views via contrastive learning. This combination yields semantic-rich trajectory embeddings capable of supporting destination prediction, arrival time estimation, and similar trajectory search with high efficiency. The approach achieves state-of-the-art or competitive results on two real-world datasets while maintaining light embedding-time and model-size, thanks to the low-cost Traj-SSM and the targetted pre-training. The work's significance lies in delivering cross-task, semantically informed trajectory representations suitable for real-time systems and large-scale analytics without prohibitive computation.
Abstract
Vehicle trajectories provide crucial movement information for various real-world applications. To better utilize vehicle trajectories, it is essential to develop a trajectory learning approach that can effectively and efficiently extract rich semantic information, including movement behavior and travel purposes, to support accurate downstream applications. However, creating such an approach presents two significant challenges. First, movement behavior are inherently spatio-temporally continuous, making them difficult to extract efficiently from irregular and discrete trajectory points. Second, travel purposes are related to the functionalities of areas and road segments traversed by vehicles. These functionalities are not available from the raw spatio-temporal trajectory features and are hard to extract directly from complex textual features associated with these areas and road segments. To address these challenges, we propose PTrajM, a novel method capable of efficient and semantic-rich vehicle trajectory learning. To support efficient modeling of movement behavior, we introduce Trajectory-Mamba as the learnable model of PTrajM, which effectively extracts continuous movement behavior while being more computationally efficient than existing structures. To facilitate efficient extraction of travel purposes, we propose a travel purpose-aware pre-training procedure, which enables PTrajM to discern the travel purposes of trajectories without additional computational resources during its embedding process. Extensive experiments on two real-world datasets and comparisons with several state-of-the-art trajectory learning methods demonstrate the effectiveness of PTrajM. Code is available at https://anonymous.4open.science/r/PTrajM-C973.