It's All in the Mix : Technology Choice between Driverless and Human-Driven Vehicles in Sharing Systems

Problem definition: Operators of vehicle sharing systems such as carsharing or ride-hailing might benefit from integrating autonomous vehicles into their fleet. Here, the impact of optimal fleet size and composition on an operator's profitability is not trivial due to the tradeoff between operational benefits and higher investment costs for driverless vehicles.Methodology/Results: We analyze a two-stage fleet sizing and composition problem. The strategic first-stage problem determines fleet size and composition decisions. The operational second stage is a rebalancing problem, formalized as a semi-Markov decision problem. We approximate the second-stage problem by a closed queueing network to devise a scalable linear programming solution approach. A numerical study on artificial and real-world instances reveals a significant profit improvement potential of driverless and mixed fleets. With equal contribution margins for both vehicle types, the profit improvement of using a driverless or mixed fleet is up to 3.9% in the real-world instances. Mixed fleets can be optimal if contribution margins are equal for both vehicle types, and become more likely optimal if the total demand increases. In mixed fleets, the number of human-driven vehicles exceeds that of driverless vehicles; on average 19.8% of all vehicles are driverless.Managerial Implications: Vehicle sharing systems can benefit from driverless vehicles. Frequently, a mix of driverless and human-driven vehicles further improves profits. Even if demand is only slightly imbalanced and contribution margins for both vehicle types are equal, operators can benefit from introducing driverless vehicles into their fleet. In many instances, using both vehicle types simultaneously increases profits. Further, once driverless vehicles are in use, operators can enter new markets that were previously unprofitable