Improving the efficiency of last-mile delivery with the flexible drones traveling salesman problem
Shih-Hao Lu,R.J. Kuo,Yi-Ting Ho,Anh-Tu Nguyen
DOI: https://doi.org/10.1016/j.eswa.2022.118351
IF: 8.5
2022-12-15
Expert Systems with Applications
Abstract:This paper analyzes how last-mile delivery performance can be improved based on combining one truck and multiple drones. We propose a Flexible Drones Traveling Salesman Problem (FDTSP) model to minimize the total delivery time. In FDTSP, a truck acts as a mobile launch platform for drones, delivers parcels to customers, and replaces drone deliveries when necessary, while drones distribute parcels to as many customers as possible to minimize the time spent on the whole route. To solve the FDTSP, we prose a two-stage heuristic formulated on bisecting K-means and metaheuristics approaches. Alternative delivery solutions for exceptional customer nodes, which exceed the drone endurance, are also introduced to ensure the deliveries for all customers in the network. The comparison analysis of three algorithms is also examined. The FDTSP significantly achieves time saving between 7.13% and 24.17% and improves the on-time service level between 13.01% and 19.77% compared to the truck-only mode. The number of customers served by drones can increase up to 56.47% if using one drone; however, additional drones may have declining marginal returns. The FDTSP reduces total tour time for last-mile delivery in urban areas with a high density of customers. Additionally, the longer the drone endurance and the faster the drone speed, the higher the efficiency of the FDTSP.
computer science, artificial intelligence,engineering, electrical & electronic,operations research & management science