The field of traffic state estimation and forecasting is rapidly evolving, with a focus on developing innovative methods to improve the accuracy and reliability of traffic predictions. Recent research has explored the use of physics-informed deep operator networks, graph neural networks, and probabilistic forecasting models to estimate traffic states and forecast traffic conditions. These approaches have shown promising results in addressing the challenges of traffic forecasting, particularly in regions without traffic observations. The integration of external knowledge, such as weather conditions and road sensor data, has also been found to enhance the generalizability of traffic forecasting models. Noteworthy papers in this area include:

• A Physics-informed Deep Operator for Real-Time Freeway Traffic State Estimation, which proposes a novel approach to traffic state estimation using physics-informed deep operator networks.
• Generalising Traffic Forecasting to Regions without Traffic Observations, which introduces a model that exploits external knowledge to compensate for missing observations and enhance generalisation.
• NEXICA: Discovering Road Traffic Causality, which presents an algorithm to discover causal relationships between different parts of the highway system.
• Probabilistic Forecasting Method for Offshore Wind Farm Cluster under Typhoon Conditions, which proposes a score-based conditional diffusion model for probabilistic forecasting of offshore wind power during typhoon events.