Sub-field · 6 papers
Physics-Informed Deep Learning Traffic
Traffic state estimation using physics-informed deep learning (PIDL) integrates physical traffic flow models—such as LWR, CTM, and second-order models—with neural networks to improve accuracy and interpretability. These works address the challenge of reconstructing spatiotemporal traffic variables (speed, density, flow) from sparse sensor data by embedding traffic flow equations as constraints during model training. A recurring focus is on hybrid paradigms that leverage computational graphs and neural operators to make the approach scalable and applicable to fundamental diagram estimation and mean field game formulations.
Papers in this community
- Physics-Informed Deep Learning for Traffic State Estimation: A Hybrid Paradigm Informed By Second-Order Traffic Models 2021 DOI ↗
- Physics-Informed Deep Learning for Traffic State Estimation: A Survey and the Outlook 2023 DOI ↗
- Physics-Informed Deep Learning for Traffic State Estimation: Illustrations With LWR and CTM Models 2022 DOI ↗
- A Physics-Informed Deep Learning Paradigm for Traffic State and Fundamental Diagram Estimation 2022 DOI ↗
- Physics-informed deep learning for traffic state estimation based on the traffic flow model and computational graph method 2024 DOI ↗
- Scalable Learning for Spatiotemporal Mean Field Games Using Physics-Informed Neural Operator 2024 DOI ↗