A Simulated annealing algorithm for the vehicle routing problem with drones
Résumé
Civil drones have been used since many years ago in different sectors, e.g., energy, agriculture,
surveillance, and emergency response. However, in recent years, many commercial logistic
companies such as Amazon, DHL, DPDgroup, etc. have started conducting experiments with
the objective of using drones in last-mile logistics, i.e., delivering parcels to the customers
[1, 4, 6, 7].
In fact, traditional package delivery systems use conventional transport vehicles, e.g., trucks,
cars, or motorcycles for delivering parcels. However, due to growing popularity of e-commerce
and importance of fast delivery in satisfaction of the customers, drones attracted attention
as an alternative and novel mode of delivery. In comparison to the usual modes of delivery,
e.g., trucks, drones might be less reliable but more cost-efficient. Furthermore, drones are
considered as a faster mode of delivery as they are not restricted to the road network, traffic
jam, and red lights. Consequently, drones are also useful in delivering parcel in areas that are
difficult to reach. Finally, drones consume less energy than trucks; hence, drones might be
considered as more environment-friendly delivering tools.
Following the practical motivations, many researchers started examining integration of delivery by drones in the classical models, e.g., traveling salesman problem with drone (TSP-D)
[1, 2] and vehicle routing problem with drones (VRPD) [6, 7, 9].
Drones are not the sole novelty of technology in parcel delivery. Automated ground vehicles
(AGV) or autonomous robots are also considered as a new mode of parcel delivery. Indeed,
robots have almost the same advantages as drones except the fact that robots are slower than
drones; however, robots are safer and more cost and energy efficient than drones. Consequently,
robots might be considered as an even more reliable delivery mode in dense urban areas [3, 5, 8].
In this study, we are interested in a variant of the Vehicle Routing Problem (VRP) in which,
in addition to trucks, drones with limited capacity are also used to bring parcels to customers.
More precisely, the drones assist the trucks in serving customers through so-called sorties. This
framework that combines trucks and drones is called the Vehicle Routing Problem with Drones
(VRPD) [7, 9]. Moreover, we investigate the ecological impact of using drones by testing energy
consumption of the truck-drone fleet.