| Project Detail |
The Online Food Delivery (OFD) industry has experienced rapid growth, largely driven by third-party platforms such as UberEats, DoorDash, and Grubhub. Central to the efficiency of these platforms is the Food Delivery Dispatch Problem (FDDP), an optimization problem that aims to match couriers with orders in real time, balancing delivery speed and costs. Current methods to solve this problem rely heavily on general-purpose optimization tools that do not fully exploit the structural properties of the underlying problem. This project, GraphEats, aims to bridge this gap by integrating advanced graph-theoretic methods, pushing the boundaries of logistics optimization in the OFD domain. GraphEats approaches the FDDP as a graph theory problem on an appropriately constructed graph, referred to as the conflict graph. In this graph, vertices represent potential plans (courier-order assignments) and edges represent conflicts between plans. Despite their pivotal role, there is currently no systematic academic study on the conflict graphs emerging in food delivery applications. Leveraging structural properties of the conflict graph, GraphEats will develop innovative, tailored preprocessing methods and algorithms for solving the FDDP, improving both the efficiency and running time of dispatch algorithms. Additionally, the project will explore how Graph Neural Networks (GNN)-based techniques can be customized for food delivery logistics by identifying meaningful structural patterns in the problem to further boost solver performance. GraphEats seeks to deliver both immediate improvements to current dispatch technology and long-term advancements in the application of graph theory to real-world logistics problems. Ultimately, this research will establish a versatile framework adaptable to a variety of logistics challenges, advancing both academic knowledge and practical solutions for the fast-growing OFD industry. |
