The MobilityLab develops data-driven AI methods for the optimization of transportation networks. Spanning over many areas of the transportation sector, the trend to more networked systems has become visible. The more details are integrated, the more complex become data analysis, control and optimization, leading to many unresolved problems. Building upon recent topics such as e-mobility, free flight and digitalization, the MobilityLab seeks to drive innovation by formulating and investigating mathematical key problems. These are the basis for realistic models and efficient algorithms, which evolve from identification, analysis and exploitation of specific problem structures. Since many structures are similar in the world of transportation, results ideally carry over to several applications, and as well to other types of network problems outside transportation.
In the context of the second phase of the Research Campus MODAL, we work on three projects in the areas of flight planning, air cargo logistics and electric bus scheduling. Our collaboration partners are FU Berlin, Lufthansa Systems GmbH & Co. KG, Ab Ovo Deutschland GmbH and LBW Optimization GmbH.
In our flight planning project, we construct dynamic super-fast shortest-path algorithms for the free flight paradigm, also in the Pareto sense concerning, e.g., flight time, fuel consumption and overflight costs. Further aspects are integrated planning and parallelization.
Duty scheduling for managing cargo containers at airports is the core of the second project. We want to find a combination of integer programming, robust optimization, stochastic methods and machine learning in order to improve the current heuristic-based state-of-the-art. For a realistic evaluation, we will also create a customizable simulation environment.
Our third project is concerned with the operation of electric buses in public transport. The limited battery ranges impose new difficulties on scheduling. Invoking machine learning methods, we propagate state of charge functions to obtain a fine-grained battery model. Moreover, we want to compute robust vehicle schedules for mixed electric and conventional fleets, incorporating our expertise on railway rolling stock rotation planning and integrated duty and vehicle scheduling.