Bridging Traffic State and Trajectory for Dynamic Road Network and Trajectory Representation Learning
Chengkai Han, Jingyuan Wang, Yongyao Wang, Xie Yu, Hao Lin, Chao Li, Junjie Wu
TL;DR
TRACK introduces a joint framework for dynamic road network and trajectory representation learning by bridging traffic state and trajectory data. It combines a trajectory-transition-aware GAT for segment dynamics, a traffic transformer encoder for bidirectional spatial-temporal traffic states, and a co-attentional transformer with a trajectory-traffic state matching objective to enable cross-view interaction, all trained with a unified self-supervised loss. The approach yields $d$-dimensional road segment representations $\boldsymbol{h}_{v,t}$ and trajectory representations $\boldsymbol{l}_{\mathcal{T}}$ that improve downstream tasks such as multi-step traffic state prediction and travel time estimation, as demonstrated on two real-city datasets with strong empirical gains and interpretable dynamics. The results, including detailed ablations and case studies, validate the effectiveness of dynamic and cross-view modeling for urban traffic analytics and provide a scalable blueprint for integrated traffic state and trajectory representation learning.
Abstract
Effective urban traffic management is vital for sustainable city development, relying on intelligent systems with machine learning tasks such as traffic flow prediction and travel time estimation. Traditional approaches usually focus on static road network and trajectory representation learning, and overlook the dynamic nature of traffic states and trajectories, which is crucial for downstream tasks. To address this gap, we propose TRACK, a novel framework to bridge traffic state and trajectory data for dynamic road network and trajectory representation learning. TRACK leverages graph attention networks (GAT) to encode static and spatial road segment features, and introduces a transformer-based model for trajectory representation learning. By incorporating transition probabilities from trajectory data into GAT attention weights, TRACK captures dynamic spatial features of road segments. Meanwhile, TRACK designs a traffic transformer encoder to capture the spatial-temporal dynamics of road segments from traffic state data. To further enhance dynamic representations, TRACK proposes a co-attentional transformer encoder and a trajectory-traffic state matching task. Extensive experiments on real-life urban traffic datasets demonstrate the superiority of TRACK over state-of-the-art baselines. Case studies confirm TRACK's ability to capture spatial-temporal dynamics effectively.