Single Transfer Journeys in Dynamic Taxi Sharing

Over the course of the last years, commercial mobility providers introduced taxi sharing services like UberXShare, managing dynamic fleets of vehicles and allow for more flexible and cheaper solutions than traditional taxis do. We propose an algorithm that computes single-transfer journeys for the dynamic taxi sharing problem. When taking on a single-transfer journey, the passenger is allowed to transfer between vehicles exactly once. Therefore, we extend the state-of-the-art dynamic taxi sharing dispatcher KaRRi. We show that our algorithm improves the no-transfer solution for up to 16.8% of the requests. We manage to reduce the total vehicle operation time by up to 5.7% while providing similar service quality for passengers at the expanse of increased running time. The running time of our algorithm
drastically increases due to the combinatorial explosion of possible assignments, that now have to consider two vehicles with feasible pickups and dropoffs as well as many possible transfer points. Providing a framework for algorithms computing single-transfer journeys, our work serves as the basis of future work for efficient filter heuristics and advanced transfer point calculation techniques.