Ph.D. Proposal

Lloyd Teta

(Professor Dimitri Mavris)

"A Framework for Concurrent Design and Fleet Operations Optimization of Unmanned Aerial Vehicle based Last-Mile Delivery Networks"

Wednesday, December 3

12:00 p.m.

Weber SST 11, Collaborative Visualization Environment  CoVE

Abstract: 

The rapid expansion of e-commerce and demand for on-demand shipping have made last-mile delivery the most expensive and logistically complex part of the supply chain. Unmanned Aerial Vehicles (UAVs) offer a promising solution to reduce costs, lower emissions, and bypass urban congestion. Deploying UAVs at scale, however, introduces unique operational challenges. Drone performance, especially range and endurance, is highly sensitive to mission-specific factors such as payload weight, route complexity required for obstacle avoidance, and variable weather conditions. This sensitivity creates a strong interdependence between the vehicle performance and the operational network requirements. Despite this interdependence, current UAV-based delivery systems are designed using a sequential design process: design the vehicle first, then optimize its operations. This approach, inherited from traditional aerospace practice, overlooks critical trade-offs inherent in drone operations due to their mission sensitivity. The result is often suboptimal delivery systems that are inefficient and costly.

This dissertation addresses this gap by developing an integrated optimization framework that concurrently solves Conceptual Vehicle Design (CVD) and Fleet-Level Operation (FLO) decisions of UAV-based last-mile delivery networks. The framework couples a physics-based CVD module, which models vehicle characteristics, with a holistic FLO model formulated as a Location-Fleet-Routing Problem (LFRP). The FLO model simultaneously optimizes strategic depot placement, tactical fleet sizing, and operational routing. A multidisciplinary design optimization (MDO) architecture manages the bi-directional information exchange between the CVD and FLO modules, which enables the system to converge on a synergistic co-design.

Tested through a case study of a delivery network based on the Atlanta region, the thesis addresses two questions: (1) quantifies the cross-level interactions between depot placement, fleet sizing, and routing and their influence on fleet delivery performance (cost and service level) and (2) explores how UAV design variables influence optimal delivery network configuration and in turn, how fleet-level operation strategy impacts the required vehicle design. Expected contributions include quantitative evidence of cross-level trade-offs and a computational framework capable of co-optimizing vehicle design and operations. The framework provides a foundation for designing more scalable, energy-efficient and economically viable UAV delivery systems that bridge engineering design and logistics planning.

Committee:

Professor Dimitri Mavris (advisor), School of Aerospace Engineering
Professor Daniel P. Schrage , School of Aerospace Engineering
Professor Kai A. James , School of Aerospace Engineering
Professor. Raphaël Gautier , School of Aerospace Engineering
Professor Marco Ceze, Amazon - Prime Air