Charging Cost-Aware Fleet Management for Shared On-Demand Green Logistic System

With the advancement of transportation electrification and Internet of Things (IoT), the cloud-based shared on-demand logistic fleet management platforms, such as Lalamove and Gogovan, become more and more popular. Under this setting, the scheduling platform needs to dispatch vehicles while dealing with the increasing charging demands of the logistic fleet and serving the dynamically arrived logistic requests. Based on the information obtained via IoT technology, the platform shall coordinate the fleet management decisions, such as logistic order matching, vehicle routing, and charging decisions to optimize the operational profit of the fleet. To solve the fleet management problem in the shared on-demand green logistic system in an online manner, we propose a deep reinforcement learning-based scheduling method. We use the real-world data to model the stochastic arriving of logistic requests and conduct experiments to explore the navigation effect of different charging pricing schemes. The simulation results demonstrate that the proposed method can adaptively optimize the fleet management decisions of the logistic fleet while enabling multiple pickups operation and maintaining the quality of delivery service.

Automated summary

With the advancement of transportation electrification and IoT, cloud-based shared on-demand logistic fleet management platforms are becoming increasingly popular. In this setting, the scheduling platform needs to dispatch vehicles while dealing with the charging demands and logistic requests. Using IoT technology, the platform coordinates fleet management decisions to optimize operational profit. To solve the fleet management problem, we propose a deep reinforcement learning-based method that adapts to the stochastic arriving of logistic requests and explores different charging pricing schemes. Simulation results demonstrate its effectiveness in optimizing decisions and maintaining delivery service quality.