The paper introduces DeliverAI, a reinforcement learning-based algorithm for food delivery. The current model of delivery is inefficient as each delivery is individually optimized. This research proposes a multi-objective optimization model that aims to enhance consumer satisfaction and reduce delivery costs. By taking inspiration from the taxi industry’s ride-sharing success, the authors developed DeliverAI, which uses path-sharing among deliveries to decrease the total distance traveled, thus reducing costs. The system was tested in a simulated environment using real data from Chicago, showing a 12% reduction in fleet size, a 13% reduction in distance traveled, and a 50% increase in fleet utilization.
Publication date: 3 Nov 2023
Project Page: https://arxiv.org/abs/2311.02017v1
Paper: https://arxiv.org/pdf/2311.02017
