An Adaptive Large Neighborhood Search for Routing and Scheduling Carsharing Service Requests
One possibility of coordinating service requests that arise for vehicles of a carsharing fleet is to optimize routes of shuttles that drop off and pick up service agents. This scenario is modeled as a variant of Vehicle Routing Problem (VRP), including aspects of the VRP with Time Windows, the Team Orienteering Problem and the Pick-Up and Delivery Problem. A metaheuristic, an Adaptive Large Neighborhood Search is adapted, tested by applying real-world data and evaluated regarding performance and run time. The results show that, despite high run times to improve the initial solution several times, a feasible solution is obtained quickly. Some very practicable routes are obtained when including the minimization of the latest arrival time in the hierarchical objective function. Then, all shuttles are occupied evenly and results reach a high number of served requests. The algorithm can support fleet managers to handle a complex problem within their daily business.
KeywordsAdaptive large neighborhood search Metaheuristic Pick-up and delivery problem Carsharing service requests
- 1.bcs. Bundesverband CarSharing. https://carsharing.de/alles-ueber-carsharing/carsharing-zahlen/aktuelle-zahlen-daten-zum-carsharing-deutschland