Learning to optimize the dispatch time interval for on-demand food delivery service

• Main Authors: YANG, Jingfeng, ZHANG, Zhiqin, LAU, Hoong Chuin
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2024
• Subjects: Intelligent Logistics; Simulation and Modeling; Other Theories; Applications; Technologies; Artificial Intelligence and Robotics; Transportation
• Online Access: https://ink.library.smu.edu.sg/sis_research/10089
• Institution: Singapore Management University

In recent years, the rapid advancement of mobile and wireless communication technologies has enabled real-time connectivity for on-demand delivery platforms, facilitating efficient door-to-door services like online food delivery. This study addresses a practical challenge faced by a food delivery platform, where customer orders must be allocated to drivers responsible for collecting food from designated centers and delivering it to customers within specific time windows. This dynamic pickup and delivery problem emphasizes prompt delivery as the critical objective. Our research focuses on optimizing the dispatch intervals for orders on such platforms. We tackle this by formulating the problem as a Markov decision process (MDP) and introducing a two-stage framework that combines a multi-agent reinforcement learning (RL) approach for order dispatching with a heuristic method for driver routing. The RL algorithm determines the optimal timing for each order's entry into the matching pool, while the routing method integrates orders into drivers' delivery routes. Extensive experiments, using real-world data and a simulator, show our results surpass benchmark methods, enhancing the efficiency of order dispatching in on-demand food delivery services.