DynaOD: Dynamic Origin-Destination Flow Generation with Discrete-to-Continuous Temporal Semantic Modeling

2026-06-08

AI Summary

This work proposes DynaOD, a novel framework for generating dynamic origin-destination (OD) flows in scenarios lacking historical OD observations. DynaOD is the first to jointly model discrete directional trends and continuous temporal evolution, producing spatiotemporally coherent OD flows that preserve urban spatial heterogeneity. The approach leverages lightweight time-varying regional representations to conditionally guide a pre-trained static OD generator, resulting in a modular and plug-and-play architecture amenable to cross-city transfer and efficient deployment. Extensive experiments on multiple large-scale real-world datasets demonstrate that DynaOD significantly outperforms existing baselines in both predictive accuracy and distributional fidelity.

Abstract

Dynamic origin-destination (OD) flow generation seeks to synthesize realistic mobility dynamics from temporal context alone, without relying on historical OD observations. A key challenge is to translate semantic temporal signals into temporally coherent OD patterns while preserving the inherent spatial heterogeneity of urban regions. We propose DynaOD, a semantic-driven framework that models temporal dynamics through two complementary perspectives: discrete directional trends that characterize qualitative shifts in urban activity patterns, and continuous temporal evolution that captures how such shifts unfold over time. By jointly encoding these temporal semantics, the framework constructs time-varying region representations that condition pretrained static OD generators in a lightweight and plug-and-play fashion. This modular design further supports scalable deployment and cross-city transferability. Extensive experiments on large-scale real-world datasets show that our method consistently outperforms representative baselines in both predictive accuracy and distributional fidelity. Code is publicly available at https://github.com/csjiezhao/DynaOD.

Problem

• dynamic origin-destination flow
• temporal semantic modeling
• spatial heterogeneity
• mobility generation
• temporal coherence

Innovation

• dynamic origin-destination flow
• temporal semantic modeling
• discrete-to-continuous modeling
• plug-and-play generation
• cross-city transferability