Computational approaches for solving two-echelon vehicle and UAV routing problems for post-disaster humanitarian operations

Abstract

During and after Hurricane Maria in 2017, the island of Puerto Rico faced an impossible situation: the transportation, power, and communication networks were all damaged, leading to numerous people being trapped. Humanitarian logistics service providers have the responsibility to reach as many affected populations as possible within short periods of time. Their extremely difficult operations are further hindered by these failures in our built environment. Failures in transportation mean that some routes cannot be used by conventional ground vehicles; failures in communication mean that demands are uncertain, and the locations of said demands may be difficult to pinpoint. In this work, we propose a novel two-echelon vehicle routing framework for performing many of the humanitarian logistics operations using aerial uncrewed autonomous vehicles (drones) to address the issues associated with these infrastructure failures. In our proposed framework, we assume that ground vehicles cannot always reach the trapped population directly, but they can transport drones from a depot to some intermediate satellite locations. The drones launched from these locations can serve to both identify demands for medical and other aids (e.g., epi-pens, medical supplies, dry food, water) and make deliveries to satisfy them. Specifically, we present an expert decision support system, in which the resulting optimization problem is formulated as a two-echelon vehicle routing problem with trucks as the first echelon vehicles and drones as the second echelon vehicles. Our framework utilizes two types of drones. Hotspot or communication providing drones have the capability of providing cell phone and internet reception, and hence, are used to capture demands and their locations. Delivery drones are subsequently employed to satisfy these obtained demands. To handle demand uncertainty, we decompose the decision problem into two stages: providing telecommunications capabilities in the first stage, thereby capturing demands precisely, and satisfying the resulting demands in the second stage. To solve the resulting models, we propose a matheuristic computational approach in the form of a decomposition algorithm inspired by column generation to identify optimal drone routes. To showcase the application of our proposed framework, we present results from numerical experiments on datasets created to simulate the demand for medical aid in Puerto Rico after Hurricane Maria.