Towards a Modular Vehicle Routing Solver framework

at  KU Leuven

TOWARDS A MODULAR VEHICLE ROUTING SOLVER FRAMEWORK

(ref. BAP-2025-45)
Laatst aangepast: 22/01/25
Within the NUMA research unit of the Department of Computer Science at KU Leuven, CODeS research group focuses on combinatorial optimization.
The overall research theme of CODeS includes the design, analysis and application of algorithms for combinatorial optimisation problems. The group investigates the construction of models, the behaviour and the application of algorithms for combinatorial optimisation and builds upon more than a decade of intense activity in this field. The group, located in Ghent, is part of the Faculty of Engineering Technology, which has a prolific industrial collaboration history. In the past few years, Reuters has consistently ranked KU Leuven as Europe’s most innovative university.
Website van de eenheid

PROFILE

We are looking for a highly motivated postdoctoral researcher, holding a degree in computer science, (applied) mathematics, mathematical engineering or related subjects.

The candidate should have proven experience in

• modelling VRP or other complex combinatorial optimization problems,
• developing and implementing efficient optimization algorithms,
• setting up and implementing a software system in a modern programming language.

The candidate should have excellent proficiency in English, both oral and written.
We expect candidates to provide evidence of these requirements.

Logistics companies and other service providers face increasingly complex vehicle routing problems (VRPs) as they are confronted with challenges stemming from e-commerce growth, just-in-time delivery demands and regulatory restrictions on working hours. Several Flemish and international companies testified how their VRPs are unique, incorporating specific business constraints or objectives that prevent them from using off-the-shelf optimization software. Recent PhD theses within the research group resulted in state-of-the-art optimization algorithms for various types of VRPs and VRP subproblems. In this project, we intend to restructure these research results into the components of a unified, configurable and extendable VRP solver software framework. The resulting framework will enable rapid customization, development and deployment of a wide range of VRP algorithms. At the same time, the framework will serve as a solid foundation for future collaborations aimed at addressing emerging VRP variants.