One Plus One Is Greater than Two for Multiple Ridesharing Service Providers: A Theoretical Proof and an Algorithm for Sustainable Collaborative Ridesharing

Ridesharing is one type of sustainable transport mode that provides a promising approach to achieving the Sustainable Development Goals (SDGs). Although users can help mitigate CO2 emissions and fuel consumption by using the services of a single ridesharing service provider, their requests may not be accepted because the service provider may not have a sufficient number of drivers to meet their requirements. The existence of multiple ridesharing service providers creates opportunities to satisfy more users’ requirements and enhance the sustainability of ridesharing services through collaboration. In one of our previous studies, we developed several metaheuristic algorithms to show the benefits of collaboration among multiple ridesharing service providers by comparing their performance with that of multiple ridesharing service providers operating independently without collaboration. However, the benefits of collaboration among multiple ridesharing service providers are demonstrated based on numerical results. There is currently no theoretical proof showing that collaboration among multiple ridesharing service providers always performs at least as well as independent operation. In addition, our previous study indicates that the development of a more effective algorithm is key to benefiting from collaboration among multiple ridesharing service providers. The goals of this study are twofold: (1) to develop a theory proving that the performance of multiple ridesharing service providers operating collaboratively is at least as high as that of multiple ridesharing service providers operating independently without collaboration and (2) to develop a more effective algorithm to further improve performance. The theory developed in this paper provides a formal proof of the benefits of collaboration among multiple ridesharing service providers for improving sustainability. The advanced algorithm proposed in this paper outperforms the 12 algorithms developed in our previous study.