Abstract
•A novel mobility-preference-based mechanism with baseline price control (MPMBPC) is designed to address on-demand first-mile ridesharing.•We prove four important properties of MPMBPC, including “individual rationality”, “incentive compatibility”, “price controllability”, and “detour discounting reasonability”.•A computationally efficient heuristic algorithm called solution pooling approach (SPA) is developed to solve large-scale ridesharing mechanism design problems.•Numerical examples demonstrate the suitability of MPMBPC mechanism and SPA algorithm for ridesharing system design.
Ridesharing emerges as a viable way to bridge the first-mile accessibility gap to public transit. There is particularly a high demand for the first-mile ridesharing service in transit-intensive metropolitan areas. This paper studies the mechanism design problem in order to promote passengers’ participation in the on-demand first-mile ridesharing accounting for passengers’ mobility preferences, including passengers’ requirements on arrival deadlines, maximum willing-to-pay prices, and detour tolerances. A rolling horizon planning approach is used to process spontaneous on-demand passenger requests. We propose a novel mechanism, namely “Mobility-Preference-Based Mechanism with Baseline Price Control” (MPMBPC), which adapts the traditional Vickrey–Clarke–Groves (VCG) mechanism and incorporates a baseline price control component. MPMBPC is proved to satisfy several important mechanism design properties, including “individual rationality”, “incentive compatibility”, “price controllability”, and “detour discounting reasonability”. In comparison with the traditional general-purpose VCG mechanism, MPMBPC can avoid unreasonably low prices and prevent carriers’ deficits. A computationally efficient heuristic algorithm called Solution Pooling Approach (SPA) is developed to solve large-scale ridesharing mechanism design problems. Numerical examples are developed to demonstrate that SPA can solve large-scale ridesharing mechanism design problems in a computationally efficient way, with satisfactory solution qualities.