Graduation Year

2025

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Industrial and Management Systems Engineering

Major Professor

Hadi Gard, Ph.D.

Committee Member

Changhyun Kwon, Ph.D.

Committee Member

Ankit Shah, Ph.D.

Committee Member

Yu Zhang, Ph.D.

Committee Member

Reza Ebrahimi, Ph.D.

Keywords

Deep Learning, Equitable Workload Allocation, Genetic Algorithm, Truck-Drone Routing, Vehicle Routing Problem

Abstract

This dissertation addresses large-scale optimization problems in transportation emerging from hierarchical decision-making, equitable workload allocation, and innovative routing logistics. It presents three sets of contributions, each detailed in a separate chapter, and offers computational tools and insights to advance both the theory and practice of transportation systems.

The first work introduces a deep learning-enhanced genetic algorithm framework for solving the Hierarchical Vehicle Routing Problems (HVRPs). Traditional optimization approaches to such problems require extensive evaluation of multiple lower-level routing solutions and are computationally intensive. Our innovative method integrates a genetic algorithm with a pretrained graph neural network, which is trained on solutions from the HGS CVRP solver, to effectively predict vehicle routing costs resulting from upper level decisions without explicitly solving each scenario. Extensive computational experiments validate the effectiveness of this approach, demonstrating significant computational savings while maintaining high-quality solutions across various datasets.

The second work proposes a practical equity-oriented optimization framework tailored specifically for crowdshipping platforms within the last-mile delivery systems. Recognizing that equitable workload distribution and fair compensation are crucial for maintaining a sustainable pool of crowdworkers, our framework utilizes bi-objective optimization to balance equity against operational costs. To effectively solve this problem, we propose Hybrid NSGA-II. Numerous experiments are conducted to quantify the trade-offs between cost and equity, revealing that modest sacrifices in efficiency can lead to substantial equity improvements. Our findings provide actionable insights for gig-economy platforms, highlighting optimal strategies to maximize fairness with minimal cost increases, and guiding workforce management by analyzing the effects of varying crowdshipper pool sizes.

The third problem extends the classical Rural Postman Problem (RPP) to a mixed-fleet scenario involving multiple trucks and drones, with the objective of minimizing makespan. Trucks act as mobile depots, deploying drones to service required arcs while also serving the arcs they visit themselves. To navigate the complexity inherent in this problem, we introduce a Hybrid Genetic Algorithm with a chromosome structure that encodes required arc sequences and vehicle assignments, accompanied by multiple dedicated local searches. Comprehensive evaluations on large-scale instances demonstrate the scalability, effectiveness, and real-time applicability of our method, underscoring significant improvements achievable through truck-drone integration in arc-routing logistics.

Download

Share

COinS