Multi-Objective Optimization for Drone Delivery

TL;DR

This paper introduces a multi-objective, stochastic optimization model for drone delivery scheduling that accounts for multiple goals and unexpected events, using real-world data for evaluation.

Contribution

It presents the first multi-objective, three-stage stochastic model for drone delivery scheduling incorporating real-world uncertainties.

Findings

Effective multi-objective optimization achieved with ε-constraint method.

Model validated using real Singapore delivery data.

Addresses unexpected drone failures impacting scheduling.

Abstract

Recently, an unmanned aerial vehicle (UAV), as known as drone, has become an alternative means of package delivery. Although the drone delivery scheduling has been studied in recent years, most existing models are formulated as a single objective optimization problem. However, in practice, the drone delivery scheduling has multiple objectives that the shipper has to achieve. Moreover, drone delivery typically faces with unexpected events, e.g., breakdown or unable to takeoff, that can significantly affect the scheduling problem. Therefore, in this paper, we propose a multi-objective and three-stage stochastic optimization model for the drone delivery scheduling, called multi-objective optimization for drone delivery (MODD) system. To handle the the multi-objective optimization in the MODD system, we apply $\varepsilon$-constraint method. The performance evaluation is performed by using a real dataset from Singapore delivery services.