Estimating emissions reductions with carpooling and vehicle dispatching in ridesourcing mobility

Abstract

Ride-hailing services provide on-demand transportation solutions by connecting passengers with nearby drivers through mobile applications. However, carpooling often fails to attract passengers as expected due to inefficient order-matching strategies. This study estimates emissions reductions with order matching and vehicle dispatching in ridesourcing mobility. An explainable machine learning with a hierarchical framework is constructed for arrival time prediction. Considering pick-up and drop-off locations within the expected departure time, on-demand order matching and vehicle dispatching optimization models are built to determine the minimum fleet size and efficient route planning. Real-world experiments are conducted with large-scale ridesharing orders in Beijing, China. In comparison to the current operations, a reduction of 25.25% in fleet size and a simultaneous decrease of 21.65% in pollutant emissions are achieved. Results demonstrate that carpooling and vehicle dispatching processes lead to a slight increase in passenger waiting time while enhancing the operational efficiency of ride-hailing services and reducing pollutant emissions.