We are looking for a researcher to join the Intelligent Systems for Mobility and Logistics department to develop a PhD project in the following scope:

Scope

Intelligent urban and interurban traffic management systems require complex control methods. Heterogeneous actors such as private cars, public transportation, goods delivery vehicles, micromobility vehicles including (e)bikes or (e)scooters, and pedestrians share a common urban space and transportation infrastructure, needing to be accommodated in an efficient and safe flow. Various traffic control regulations and strategies, such as traffic lights, help determine priorities and how these agents interact to reduce conflicts and bottlenecks.

The high level of digitalization, detection capabilities, and communications nowadays enable a more dynamic decision-making process. For instance, urban access regulations may include dynamic access control and restrictions for certain vehicles (such as low-emission zones or residential areas). Additionally, it also involves intersection management through Traffic Signal Control (TSC), which can adapt to unexpected incidents or congestion, prioritize public transport if it's experiencing delays, or give priority to approaching emergency vehicles at the intersection. Making decisions without prior analysis of the consequences can even lead to relieving congestion in certain areas at the expense of increasing congestion in others, thus failing to meet global objectives.

Therefore, building digital replicas of connected transportation infrastructures and moving actors can be of great assistance. In the literature, very different transportation models can be found, used by the most advanced commercial programs for transportation planning and simulation. For instance, transportation planning studies based on mesoscopic and macroscopic models assess mobility demand and support the necessary changes and investments to adjust the capacity of the transportation network infrastructure. This aims to meet current and future travel flows for medium to long-term horizons. Models can be expanded and applied to understand the temporal effects of any planned short-term intervention or unexpected incidents on overall mobility patterns. Microscopic models operate at the vehicle or actor level, providing fine-grained detail within the mobility flow. Multiple approaches are used to solve different problems, and there is no one-size-fits-all solution when it comes to building a digital replica of mobility behavior.

The objective of the PhD. will be to work on the inclusion of diverse control actuation strategies into transportation models. Diverse challenges remain unsolved in the state-of-the-art, that will be explored. For instance, since the effects of lane closures and road works in upstream and downstream traffic are not isolated and thus a global perspective of the network is sometimes needed, the inclusion of large-scale real time data into city-wide and regional decision-making processes opens scalability challenges. Moreover, the deployment of ML techniques such as reinforcement learning, well suited to solve control problems, is currently limited in real-life traffic management scenarios and intensive usage of simulation environments is required to ensure the validity of the actuation strategies.