Comparison of trip matching algorithms for mobility sharing applications

Trip matching algorithms used in ride-sharing and carpooling systems share the common goal of optimizing the number of used vehicles to satisfy a set of trips according to their temporal and spatial constraints, in order to better allocate resources and reduce traffic and congestion. However, each matching algorithm could be designed to pursue a different objective like, for instance, reducing users' waiting time for quality of service, reducing the total amount of traveled distance within the system to reduce carbon footprint, or maximizing the time two trips are shared to favor user interaction. Changing the final system objective could significantly change the performance of the system itself. In this paper, we compare the performance of matching algorithms with different objectives and show whether potential tradeoffs exist in pursuing these objectives, taking into consideration all the actors in play in a mobility sharing application. In particular, by applying the matching algorithms to two sets of daily trips performed in the cities of Pisa, Italy, and Cambridge, USA, our analysis shows that there is a matching algorithm among the ones we tested able to provide a good compromise between different optimization objectives.