Parameter-less (meta)heuristics for vehicle routing problems

Solving rich vehicle routing problems (VRPs) is a vital research topic due to their wide applicability. Although there exist various (meta)heuristics to tackle VRPs, most of them require a practitioner to tune their parameters before the execution. It is challenging in practice, since different algorithm variants often perform well for different scenarios. In this work, we present our adaptive heuristics for this task, in which we benefit from the adaptation schemes executed before the optimization. Extensive experiments backed up with statistical tests revealed that our heuristics is automatically adapted to effectively solve a given transportation problem, and retrieve routing schedules of the state-of-the-art quality.