Efficient traffic management is increasingly critical to modernizing Swedish infrastructure and reducing urban congestion. The Swedish Transport Administration constantly monitors traffic flow to ensure smooth travel and reduce delays, especially on major highways. A key component of this effort is the Motorway Control System (MCS). By continuously collecting and processing real-time traffic data, the MCS displays recommended speeds on highway gantries to harmonize traffic flow and enhance road safety. Currently, the Swedish Transport Administration leverages this system to manage active congestion. However, generating these speed recommendations relies heavily on current conditions or deterministic short-term assessments. Predicting future traffic states and anticipating sudden disruptions remains a complex challenge. Traditional forecasting methods often struggle to quantify the uncertainty of their predictions, which is essential for proactive rather than reactive traffic control. This project aims to advance spatiotemporal traffic forecasting by developing sophisticated data-driven models. By integrating the MCS data from the E4 motorway with advanced probabilistic generative modeling, the project will create a robust, forward-looking prediction framework. Specifically, this framework addresses the limitations of current deterministic estimations and predictions. By leveraging advanced machine learning architectures, such as conditional generative models, we will capture the complex spatial dependencies across the highway network. These models will generate high-resolution forecasts of traffic conditions and provide reliable confidence intervals. Ultimately, this enhanced forecasting capability will allow the MCS to post smarter, proactive speed recommendations on gantries, supporting more resilient and environment-sensitive traffic management along the E4. The main supervisor for this PhD student is Xiaoliang Ma, affiliated with the School of Architecture and the Built Environment, KTH Royal Institute of Technology.