ElasticShare: Ridesharing Order Dispatching with Dynamic Supply-demand Distribution

Shuxin Ge, Xiaobo Zhou, Tie Qiu, Guobin Wu, Xiaotong Wang

Abstract: The mobility on demand (MoD) system relieves traffic pressure by simultaneously dispatching multiple orders to a vehicle via ridesharing. However, since the supply-demand distribution varies over time, existing dispatching methods for minimum fleet failed to achieve long-term supply-demand equilibrium, and thus greatly reduces the order completion rate. In this paper, ElasticShare, a ride-sharing order dispatch method, is proposed to maximize the order completion rate under dynamic supply-demand distribution. First, we formalize the ridesharing order dispatching problem as an offline optimization problem and then prove it can be solved in an online manner by adding dummy orders and vehicles satisfying long-term supply-demand equilibrium conditions when dispatching. Second, based on the proof, we generate a certain number of dummies according to the supply-demand relation to promote or restrain ridesharing in an elastic manner. The characteristics of the dummies are determined by statistics and a specific selection method to ensure that the solution approximates the long-term supply-demand equilibrium. Next, we decouple the online problem into two sub-problems (order associating and order association dispatching), which are solved by a greedy algorithm and correctional order association dispatching algorithm, respectively. Finally, simulations with real-world data are used to validate the superiority of ElasticShare over state-of-the-art solutions in terms of the order completion rate.