TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability

2025-05-19

AI Summary

Existing GPS trajectory models exhibit poor cross-regional and cross-task transferability, necessitating frequent retraining. This paper proposes the first general trajectory learning framework supporting zero-shot/few-shot regional transfer and task-switching without retraining. Methodologically, it introduces (1) an RTTE backbone network jointly optimized with a TRIE+ spatial-contextual Mixture-of-Experts (MoE) mechanism to enhance regional generalization; and (2) a unified modality- and trajectory-point-level masked reconstruction framework enabling task-adaptive transfer. The approach integrates multimodal features—including spatial, temporal, POI, and road-network representations—alongside relative positional encoding, MoE architecture, and self-supervised pretraining. Evaluated on three real-world datasets, the framework achieves state-of-the-art performance in task transfer and improves trajectory prediction accuracy by 12.7% on average under zero-shot/few-shot regional transfer, significantly outperforming existing methods.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Vehicle GPS trajectories provide valuable movement information that supports various downstream tasks and applications. A desirable trajectory learning model should be able to transfer across regions and tasks without retraining, avoiding the need to maintain multiple specialized models and subpar performance with limited training data. However, each region has its unique spatial features and contexts, which are reflected in vehicle movement patterns and difficult to generalize. Additionally, transferring across different tasks faces technical challenges due to the varying input-output structures required for each task. Existing efforts towards transferability primarily involve learning embedding vectors for trajectories, which perform poorly in region transfer and require retraining of prediction modules for task transfer. To address these challenges, we propose TransferTraj, a vehicle GPS trajectory learning model that excels in both region and task transferability. For region transferability, we introduce RTTE as the main learnable module within TransferTraj. It integrates spatial, temporal, POI, and road network modalities of trajectories to effectively manage variations in spatial context distribution across regions. It also introduces a TRIE module for incorporating relative information of spatial features and a spatial context MoE module for handling movement patterns in diverse contexts. For task transferability, we propose a task-transferable input-output scheme that unifies the input-output structure of different tasks into the masking and recovery of modalities and trajectory points. This approach allows TransferTraj to be pre-trained once and transferred to different tasks without retraining. Extensive experiments on three real-world vehicle trajectory datasets under task transfer, zero-shot, and few-shot region transfer, validating TransferTraj's effectiveness.

Problem

• Transfer vehicle trajectory models across regions without retraining
• Enable task transferability with unified input-output structures
• Handle diverse spatial contexts and movement patterns effectively

Innovation

• RTTE module integrates multiple trajectory modalities
• TRIE module incorporates relative spatial feature information
• Task-transferable input-output scheme unifies different tasks