An adaptive large neighborhood search method for the drone–truck arc routing problem

Abstract

For applications such as traffic monitoring, infrastructure inspection, and security, ground vehicles (trucks) and unmanned aerial vehicles (drones) may collaborate to finish the task more efficiently. This paper considers an Arc Routing Problem (ARP) with a mixed fleet of a single truck and multiple homogeneous drones, called a Drone–Truck Arc Routing Problem (DT-ARP). While the truck must follow a road network, the drone can fly off of it. With a limited battery capacity, however, the drone has a length constraint, i.e., the maximum flight range. A truck driver can replace a battery for the drone after each flight trip. We first transform the DT-ARP into a node routing problem, for which we present a MIP formulation for the case with a truck and a drone. To solve large-size instances with multiple drones, a heuristic method based on Adaptive Large Neighborhood Search is proposed. The performance of ALNS is evaluated on small-size randomly generated instances and large-size undirected rural postman problem benchmark instances. In addition, an analysis is provided on the relationship between truck/drone speeds and the drone’s flight range, which affects the difficulty level to solve. The robustness of ALNS is shown via numerical experiments.