This academic article discusses a new approach to solving large-scale vehicle routing problems (VRPs) using a machine learning-based framework called decompose-route-improve (DRI). The DRI framework groups customers using clustering, taking into account their spatial, temporal, and demand data. This data-based approach outperforms traditional methods, providing high-quality solutions faster by effectively reducing complexity. The article also highlights the flexibility of DRI as it can be adapted to various VRP characteristics, making it a generalizable solution to routing problems.
Publication date: 2 Feb 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2402.00041
