Ridepooling with trip-chaining in a shared-vehicle mobility-on-demand system

This work considers an approach to uncover ridepooling opportunities in a shared-vehicle, mobility-on-demand system. In this setup, a fleet of self-driving vehicles, each capable of serving up to two passenger requests at a time, is used to transport passengers. In line with existing work, a station based model is used to represent the movement of passengers and vehicles in an urban environment. The problem of how to best route vehicles, to minimizes either (i) the total system travel time, (ii) total passenger travel time, or (iii) a weighted combination of the two, is given as the solution of a integer linear program. The formulation captures the need to rebalance vehicles from popular destination stations to popular origin stations. Moreover, it allows trips to be chained together, such that a passenger may board a vehicle with one passenger already on board. Theoretical results address the stability of such systems in terms of fleet size and characteristics of the travel demand. Finally, a simulation of a hypothetical shared mobility system, using real taxi trip data from NYC, demonstrates the efficiency gains possible by combining trips using our routing policy.