The thesis will concentrate on the analysis and automated generation of data instances based on real-world vehicle routing problems (VRPs) solved by the company Wereldo.com.

The initial steps of the thesis will analyze existing VRPs provided by the company. Based on their characteristics, a static VRP generator will be implemented using the OpenStreetMap data sources and the generator published in [2]. The static generator will then be extended to generate dynamic events, specifically arrivals of new customer requests concurrently with the routing plan execution.

The key part of the thesis will concentrate on strategies that introduce flexibility into the routing plans. These strategies will be based on forcing additional placeholder customers into the static VRP instances. Such instances would be solved with an existing solver [1] to produce a routing plan. During routing plan execution, the placeholders can be instantiated into the dynamically arriving customer requests, effectively promoting spatiotemporal flexibility in the routing.

The suitability of the placeholder-generating strategies will be evaluated with the help of the existing vehicle routing solver [1]. A strategy with no placeholders will serve as a baseline. The simulation-based experimental evaluation will concentrate on the empirical flexibility (ability to accommodate dynamic customers) and transportation costs of the routing plans produced under the different placeholder generation strategies. The thesis will publish the generator together with the representative collection of the datasets.