1. Phythics informed deep learning for traffic state estimation
Traffic state estimation (TSE) is the process of determining the current status of a traffic network, typically involving the estimation of variables such as traffic flow, speed, and density. In this work, we utilize physics-informed deep learning and generative models for TSE. We aim to improve prediction results when observations are limited. We quantify the uncertainty of the traffic state using generative models such as GANs and normalizing flows. Additionally, we are able to learn the fundamental diagram of the data.