Interactive Utility Maximization in Multi-Objective Vehicle Routing Problems: A "Decision Maker in the Loop"-Approach

Abstract

The article presents an interactive multi-criteria approach for the resolution of rich vehicle routing problems. A flexible framework was built to be able to deal with various components of general vehicle routing problems, e.g. the consideration of multiple objectives or different types of specific complex side constraints such as time windows, multiple depots or heterogeneous fleets. In the framework, a local search approach on the basis of variable neighborhood search (VNS) constructs and improves solutions in real time. The decision maker is actively involved into the resolution process as the system allows the interactive articulation of preference information, influencing the global utility function that guides the search. Results of test runs on multiple depot multi-objective vehicle routing problems with time windows are reported, simulating different types of decision maker behaviors