ORCID

https://orcid.org/0009-0009-7993-6824

Keywords

Truck-Drone Routing and Scheduling, Last-Mile Package Delivery, Vehicle Routing Problem, Combinatorial Optimization, Reinforcement Learning

Abstract

Modern delivery networks increasingly rely on innovative technologies such as drones working alongside trucks to meet rising demands for fast and efficient last–mile transportation. Hybrid truck–drone delivery offers strong potential, but its effectiveness depends on careful synchronization between trucks and drones. Although drones provide fast, direct travel, their limited endurance and the need to coordinate launch and recovery with trucks create tightly coupled routing and scheduling interactions. Poor dispatch decisions can cause truck waiting, reduce parallelism, and increase mission completion time. These challenges motivate decision frameworks that explicitly model synchronization and develop scalable methods for coordinated truck–drone operations. This dissertation develops learning– and optimization–based methods for synchronization-aware truck–drone delivery across three decision layers. Study I establishes a foundational one-truck one–drone model formulated as an episodic Markov decision process that preserves core endurance and timing constraints. Exact dynamic programming and a compact MILP are used for benchmarking, and a structured reinforcement learning policy with feasibility masking learns efficient joint decisions in this controlled setting. Study II focuses on execution–layer dispatch when the truck follows a fixed route. An event–driven simulator enforces launch, recovery, and single–operator constraints, and a masked PPO policy learns to reduce makespan consistently relative to strong greedy heuristics, especially when endurance levels make synchronization decisions critical. Study III addresses full planning in multi-truck multi–drone systems with flexible launch and recovery stops. A new integrated MILP formulation supports small–instance benchmarking, and a scalable MILP–assisted ALNS matheuristic solves large instances. Results show that coordinated planning with flexible stops provides significant reductions in makespan, particularly in larger and more spatially dispersed service regions. Together, the three studies provide a coherent toolchain for synchronization–aware decision making in hybrid truck–drone logistics.

Completion Date

2026

Semester

Spring

Committee Chair

Rabadi, Ghaith

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

School of Modeling, Simulation, and Training

Format

PDF

Document Type

Thesis

Identifier

DP0053172

Download

Share

COinS
 

Accessibility Statement

This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.