Optimal delivery route planning for a fleet of heterogeneous drones: A rescheduling-based genetic algorithm approach

Abstract

The past decade has witnessed the rapid growth of pilot studies and the adoption of drone-assisted delivery by many e-commerce companies. In practice, delivery companies usually assemble fleets of drones consisting of heterogeneous configurations to satisfy various customer package demands, and this calls for in-depth research on how to deliver packages via drones efficiently. In this study, we extend the classic vehicle routing problem (VRP) by optimally scheduling the operations in a package depot center with a fleet of drones with different capacities, speeds, and maximum flight ranges. The drones travel multiple sorties to serve customers nearby. We formulate this problem as a mixed integer programming (MIP) model that minimizes the total delivery time for a set of packages. Due to the NP-hardness of the problem, we first develop a FIFO-based genetic algorithm (FBGA) to solve the problem. In anticipation of multiple trips, we further propose a modified, rescheduling-based genetic algorithm (RBGA) that incorporates a small-scale routing-planning optimization model that “reschedules” drone operations at the end of drone sorties, with the objective of minimizing the completion time. The RBGA features the combination of the knapsack problem, the shortest path problem, and a scheduling problem, from which near-optimal delivery plans are obtained for the fleet of drones. We conduct extensive numerical experiments with different numbers of customers and drones to evaluate the performance of the algorithms. The results suggest that compared with MIP, the proposed algorithms solve the problem efficiently for the testing instances, particularly for large instances that MIP fails to solve in time, indicating the practicality of the proposed methods for real-world implementations.