OPTIMIZING DYNAMIC RIDE-SHARING SYSTEMS

Internet-enabled technologies are becoming more and more used, so that users are constantly connected to the network in every place and daily activity, and can use transportation related features, such as mobile payment systems and GPS connections, or receive real-time information on public transport timetable, traffic congestion, and so on, being then able to plan their mobility at the best. In this framework, in effect, among the innovative mobility systems implemented and developed using these technologies, the new real-time capabilities of dynamic ride-sharing, which is an extended version of the traditional ride-sharing, can play a key role if the relevant performances are improved. In other words, although ride-sharing is not a new idea, recent technological advances should increase its popularity. In this paper, a ride-sharing system is proposed whose dynamic behavior consists of two levels: the first, disaggregate level takes into account the positions and speeds of all the drivers’ vehicles, as well as the dynamics of the traffic flows in which such vehicles “move”; the second level only considers the interactions of drivers/riders with the system manager, and the interactions between drivers and riders. Then, with the aim of optimizing the performances of the above introduced RS system only the second kind of dynamics is considered, designing a pick-up and delivery optimization model able to allocate an empty seat in a vehicle to a rider through an optimal matching problem is proposed also providing, at a time, the expected performances of the system for each user. In particular, the optimization problem finds the best match and path in the considered transportation network that minimize the difference between the desired departure and arrival times, also providing, by means of suitably defined auxiliary variables, the generalized costs of each trip. The paper is organized as follows. First the ride-sharing model is introduced. Then, after the statement of the optimal matching problem and the discussion on its solution and integration with modal choice models, a simulation model is described. Finally, a real-world based case study is presented and discussed.