A multimodal mobility-on-demand service: dynamic rescheduling approach considering passenger-vehicle couplings

Abstract

Abstract Out of concern for the transport efficiency and sustainable development of city mobilities, both policymakers and commuters are calling for an integrated multimodal transportation network. Although the implementation of Mobility-on-demand (MoD) service should have been satisfying, the lack of corporation between different transit modes results in unsatisfying passenger experiences and low service efficiency. Multimodal MoD (MMOD) service is proposed to comprehensively utilize fixed-route public transit and demand-responsive transit resources for the integration of a multimodal transportation network. To formulate the MMoD, a service framework that periodically reassigns real-time orders and reschedules operational routes is proposed. To reduce the computational complexity, a passenger-vehicle routing model with multimodal transfers is constructed based on multilayer space-time networks for decoupling vehicle routes and passenger schedules. A column-generation algorithm is proposed to further reformulate, dualize, decompose and solve the primal problem of the aforementioned model. Toy example demonstrates that the proposed algorithm can yield a maximum of 60 times speed-up than that of the commercial solver Gurobi. Finally, by adjusting the passenger acceptance rate of rideshares and transfers under large-scale examples, we presented the sensitivity analysis taking the proportion of served passengers as the indicator of service efficiency. Experiments show that the passenger acceptance rate of ridesharing has a greater impact on the service efficiency.