OPTIMIZATION OF DESTINATIONS AND PATHS FOR QUADCOPTER DELIVERY OF AMAZON PARCELS

The aim of this project is to optimize a package delivery quadcopter route in a heavily-populated city. Current limitations on using quadcopters for delivering packages prevent one quadcopter from carrying more than one package in one trip. The research is motivated by an interest in investigating the economic benefits of a quadcopter carrying more than one package at a time, enabling visits to multiple destinations before returning to a base station.<br> This problem comprises two components: a path optimization to find the distance between destinations, and a destination optimization to find the optimal tour that passes through all destinations. The standard path optimization was solved using a gradient-based optimizer (MATLAB function fmincon), and a genetic algorithm was used to solve the destination optimization problem. Constraints added to the destination optimization problem, including the total weight of packages carried by a single quadcopter and the total distance of a single subtour, help make the model developed in this problem more realistic.

本项目旨在优化人口稠密城市中的包裹配送四旋翼无人机（quadcopter）航线。当前四旋翼无人机在包裹配送领域的应用限制为：单架无人机单次航程仅可搭载一件包裹。本研究的核心动机在于探究单架四旋翼无人机单次搭载多件包裹所能带来的经济效益，使其可在返回起降基站前造访多个配送目的地。
该问题包含两个核心子模块：其一为路径优化模块，用于计算各配送目的地间的通行距离；其二为目的地优化模块，用于寻找可遍历所有目的地的最优巡访航线。标准路径优化问题采用基于梯度的优化器（MATLAB函数fmincon）求解，而目的地优化问题则通过遗传算法完成求解。本次研究在目的地优化问题中加入了两项约束条件，即单架无人机搭载包裹的总重量上限与单次子航程的总距离上限，以此令本研究构建的模型更贴合现实配送场景。