UAV-aided mixed fleet management and operations for food delivery under travel time uncertainty

Rapid response time and cost-effective delivery are critical for success in the competitive food delivery market. Many food delivery service providers are integrating unmanned aerial vehicles (UAVs) into their fleets to balance cost and efficiency. However, UAVs are vulnerable to adverse weather conditions like wind and rain, impairing their performance. This not only renders UAV-equipped couriers less efficient than conventional ones but also increases costs due to specialized equipment. To address this, we propose a mixed-fleet food delivery model that optimizes the composition of UAV-equipped and conventional couriers. We formulate the problem as a two-stage stochastic program, capturing travel time uncertainties caused by weather variability. To solve large-scale instances efficiently, we develop a tailored Adaptive Large Neighborhood Search (ALNS) algorithm with problem-specific operators. Extensive experiments using real-world data demonstrate that our mixed-fleet approach outperforms homogeneous fleet strategies, i.e., all-conventional couriers or all UAV-equipped couriers, across varying uncertainty levels. Sensitivity analyses on UAV costs, fleet size, and courier expenses provide policy insights for UAV-aided mixed fleet management and operations for food delivery.