Rebalancing in Shared Mobility Systems – Competition, Feature-Based Mode Selection and Technology Choice

Vehicle sharing services commonly face the problem of unbalanced demand: The number of vehicles rented from a location may not equal the number of vehicles returned to this location. To counteract demand imbalances, operators rebalance their fleet, i.e. move vehicles from locations with an excess supply to locations with an excess demand. We investigate three extensions of the rebalancing problem: competition, modal selection, and driverless vehicles. With increasing competition, operators must consider where other operators currently have vehicles, as well as how they rebalance their fleets. We present a model that considers competition in rebalancing. In Munich, operators can gain up to 40% of their profit due to considering competition. However, operators in Munich lose up to 12% of their profit due to the presence of competition (compared to a merger). Vehicles can be rebalanced by loading them onto a truck, or by driving them. In the latter case, staff must be rebalanced too, i.e., workers have to give each other lifts, bike or use public transit to reach the next vehicle. We study which features drive the choice for either of the modes using classifiers based on multiple linear regression, multinomial logistic regression, and decision trees. With this novel approach, we show that the modal choice is mainly driven by wages for workers, and vehicle costs. The advent of driverless vehicles will directly impact the shared mobility market, as operators consider whether to procure driverless vehicles. We study the technology choice and mix problem operators face, balancing investment costs with operational costs and contribution margins. In a case study using data from Chinese ride-hailing provider DiDi, operators often benefit of mixed fleets if direct costs are equal