Optimizing the Design and Operation of Coordinated Transportation and Logistics Systems
Name
schmid-aschmid-phd-orc-2026-thesis.pdf
Size
31.42 MB
Format
Adobe PDF
Checksum (MD5)
32bf824e8bdd4ebc14a595eccde929c7
Author(s)
Schmid, Alexandria
Advisor(s)
Jacquillat, Alexandre
Date Issued
May 2026
Publisher
Massachusetts Institute of Technology
Abstract
Transportation and logistics systems are critical for efficiently moving people and goods. Amid rising customer expectations and operational challenges, transportation and logistics companies are looking for ways to improve the sustainability and reliability of their operations. Alternative operating models are emerging to tackle these issues, by leveraging telematics and real-time tracking to enable coordination and consolidation for increased efficiency. However, coordination requirements induce large-scale routing problems with complex spatiotemporal synchronization constraints, necessitating new optimization models and tailored decomposition algorithms to handle them at scale. This dissertation focuses on developing models and algorithms to enable coordination-based operating models in transportation and logistics. The first chapter focuses on optimization of relay logistics, an alternative operating model for long-haul trucking that decomposes origin-destination order routes between pit-stops to improve drivers’ lifestyle, inducing a complex coordinated routing problem. We devise a theoretical model to show that relays can facilitate load consolidation along high-volume corridors, particularly under demand imbalances and variability. To support relay operations, we formulate an integer optimization model on coupled time-space networks and develop a novel multi-arc generation algorithm to solve it at scale by building time-space networks iteratively. Computational results confirm our theoretical insights by showing that relay logistics can improve over point-to-point trucking in terms of environmental footprint, driver lifestyle, and service level. The second and third chapters focus on the rapidly growing area of robotic warehousing. Advances in computer vision, sensing, and mapping technologies have led companies to deploy large robotic fleets, requiring system-wide algorithms to manage them. In the second chapter, we develop an integer optimization model to coordinate robotic agents in parts-to-picker operations. We solve it via large-scale neighborhood search, with a novel learn-then-optimize approach to subproblem generation. In collaboration with Amazon Robotics, we show that our model and algorithm generate much stronger solutions for practical problems than state-of-the-art approaches. The third chapter develops real-time task assignment algorithms for parts-to-picker operations at the full scale of an Amazon warehouse. First, we provide theoretical results that show that coordination opportunities expand as the system scale increases. Second, we present a formulation of the task assignment problem and provide valid inequalities and principled heuristics to solve full-scale instances. Routing simulations show that the model improves the overall operational efficiency of the task assignment in a warehouse setting. The final chapter focus on the design and operation of demand-responsive deviated-fixed route microtransit systems which combine elements of public transit and ride-hailing. We formulate a two-stage stochastic optimization model that optimizes network design and service scheduling decisions in the first stage and on-demand routing deviations in the second stage, under demand uncertainty. To solve it, we develop a novel double decomposition algorithm combining Benders decomposition and column generation which outperforms all benchmarks. Results show that microtransit, supported by the double decomposition algorithm, can lead to win-win-win outcomes in urban mobility in terms of efficiency, equity, and sustainability.
MIT Department
Massachusetts Institute of Technology. Operations Research Center
Sloan School of Management
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
