The thesis is aimed at solving vehicle routing problems where we have initial routes for the vehicles and a set of new requests with changes, which must be incorporated into the existing solution in a short time frame. The routing problem inspired by the needs of the company Wereldo involves many real-life features such as heterogeneous vehicles, private and common carriers, multiple time windows, or multiple depots for vehicles.

Based on the approach implemented in the earlier master thesis [1], this work should extend the proposed methodology by the inclusion of more realistic problem features and provide high-quality solutions even for a realistic problem with many real-life features. The experimental part of the thesis will develop new problem instances with the help of the generator [1] and representative benchmark problems from Wereldo. The thesis will contain experimental analysis including parameter setting and evaluation on the generated data sets by means of the quality, runtime, and scale of the realized changes to the existing solution.