An integrated ride-matching and vehicle-rebalancing model for shared mobility on-demand services
针对共享按需出行服务，提出一个同时优化乘客匹配和车辆再平衡的集成模型，使用图方法求解，并用大规模数据验证其比传统顺序模型更能提升服务水平和可持续性。
Shared mobility on demand (MoD) services are receiving increased attention as many high-volume ride-hailing companies are offering shared services (e.g. UberPool, LyftLine) at an increasing rate. Also, the advent of autonomous vehicles (AVs) promises further operational opportunities to benefit from these developments as AVs enable a centrally operated and fully connected fleet. There are two fundamental tasks for a shared MoD service: ride-matching and vehicle-rebalancing. Traditionally, these two tasks are optimized sequentially and independently. The paper formulates an integrated ride-matching and vehicle-rebalancing problem for shared MoD services which simultaneously optimizes these two tasks. We propose a graph-based methodology to solve the integrated ride-matching and vehicle-rebalancing problem with a novel rebalancing cost term quantifying supply contributions of vehicle scheduling to zonal supply deficit balances (deviations from the desired supply level) in the network. The integrated model performance is validated using a large-scale empirical shared MoD dataset by comparing with state-of-the-art sequential models. Generally, the integrated model improves the level of service and sustainability performance compared to the sequential model. The detailed analysis shows that the vehicle rebalancing in the integrated model is replaced by a more effective ride-matching and penalizing singly served trips in the integrated model can further improve its sustainability performance.