Designing taxi ridesharing systems with shared pick-up and drop-off locations: Insights from a computational study

Taxi ridesharing (TRS) systems are considered one means towards more sustainable transportation by increasing car occupancy rates and thereby significantly improving the efficiency of urban transportation systems. In this study, we consider TRS with shared pick-up and drop-off locations, where customers of a shared trip might be required to walk a short distance from their origin/to their destination. Related research has discussed the advantages of this approach over other TRS variants, including shorter travel times, lower fuel consumption and fewer emissions. However, these studies do not investigate how a TRS ought to be designed under different environmental conditions to maximize its effectiveness in terms of rejection and sharing rate. We contribute to closing this gap with three achievements. First, we propose a new mathematical model that provides a conceptualization of the TRS problem with shared pick-up and drop-off locations. Second, we implement a rolling horizon approach and conduct extensive computational experiments based on empirical data from New York City and Porto. In our experiments, we vary and combine several exogeneous (environmental) and design-oriented factors and show that both exert considerable influence on the rejection rate, sharing rate and service quality. Third, for practitioners considering a TRS with shared pick-up and drop-off locations, our guidelines highlight the importance of system design, particularly in leveraging extended waiting times to attain low rejection rates and foster high sharing rates.