Model Results Showed that Depot-Based Drone Delivery Led to up to 60 Percent Cost Savings Compared to Truck-Only Delivery When Servicing Low Demand in Small Areas.
Delivery Route Optimization Study Compared Different Delivery Options Utilizing Drones and Trucks.
Rantoul
Analysis of Drone-based Last-mile Delivery Systems under Aerial Congestion: A Continuum Approximation Approach
Summary Information
Unmanned Aerial Vehicles (UAV), commonly known as drones, are considered as potential alternatives for freight deliveries for meeting short-range mobility needs. UAV offer great efficiency, speed, higher automation, and more direct point-to-point deliveries due to their ability to navigate low-altitude airspace. Given this view, this study presented a practical design framework to determine decisions for UAV and truck routing in a collaborative last-mile delivery system. Using an optimization model for the analysis of both truck routing and aerial-UAV traffic, the study compared the operational cost and efficiency of different delivery modes using a hypothetical numerical example. In addition, the efficiency of two different optimization solvers was compared using a real roadway network in Rantoul, Illinois.
METHODOLOGY
This study explored four delivery methods: direct drone deliveries from a depot (DD), drone deliveries from stationary trucks (STD), deliveries from moving trucks (MTD), and traditional truck deliveries (BENCHMARK). It modeled the routing of trucks and drones across regions as a complex Vehicle Routing Problem (VRP). Initially, a hypothetical city with a square grid layout and a central depot was used to estimate delivery costs for each method. The study further tested the framework on a real road network in Rantoul, Illinois, identifying six residential areas with high delivery demand and using a warehouse as the depot. The study evaluated it through both an exact solver and a heuristic algorithm. The study then illustrated the applicability of the proposed decision framework with respect to two different optimization solvers, including an exact solver and a heuristic algorithm.
FINDINGS
- The results from the hypothetical city showed that, for a small delivery region of 0.25-mile squares, the DD mode led to up to about 60 percent cost savings compared to the BENCHMARK delivery mode with trucks only (comparing $2.2 and $0.9, per delivery). For MTD mode, the savings were around 23 percent compared to BENCHMARK, and for STD, they were around 41 percent.
- The results of the comparison of two different solvers for the optimum routing strategy revealed that the exact mixed-integer program solver yielded an average delivery cost of $1.12 per parcel, and the proposed heuristics solver yielded $1.63 per parcel. It is important to note that the exact solution was computed within five minutes (or 300 seconds) with the mixed integer program technique, while the proposed heuristics solver yielded the result in 20 seconds (93 percent faster).
