Hierarchical data-driven vehicle dispatch and ride-sharing

Modern transportation system suffers from increasing passenger demand, limited vehicle supply and inefficient mobility service. Towards building an intelligent transportation system to address these issues, we propose a hierarchical framework to implement strategies that is capable of allocating vehicles to serve passengers in different locations based on limited supply. In the higher hierarchy, we optimize idle mileage induced by rebalancing vehicles across regions using receding horizon control towards current and predicted future requests. In addition, we design a dispatch strategy that is robust against passenger demand and vehicle mobility pattern uncertainties. In the lower hierarchy, within each region, pick-up and drop-off schedules for real-time requests are obtained for each vehicle by solving mixed-integer linear programs (MILP). The objective of the MILP is to minimize total mileage delay due to ride-sharing while serving as many requests as possible. We illustrate the validity of our framework via numerical simulations on taxi trip data from New York City.