Institutional Repository
The UDSpace Institutional Repository collects and disseminates research material from the University of Delaware.
- Faculty, staff, and graduate students can deposit their research material directly into UDSpace. Faculty may use UDSpace to fulfill the University of Delaware Faculty Senate Open Access Resolution, and in many cases may use it to fulfill open access requirements from grant funding agencies.
- Departments can use UDSpace to publish or distribute their working papers, technical reports, or other research material.
- UDSpace also includes all doctoral dissertations from winter 2014 forward, and all master's theses from fall 2009 forward.
To learn more about UDSpace, and how you can make your research openly accessible to the public, visit our UDSpace Policies website.
Communities in DSpace
Select a community to browse its collections.
Recent Submissions
2025, 11th Issue
(Newark, Del.: Chesapeake Pub. Corp., 2025-03-14) Newark post
LEVERAGING DEEP LEARNING FOR ROBUST VISUAL AND VISUAL-INERTIAL SLAM
Merrill, Nathaniel Wallace
Simultaneous localization and mapping (SLAM) arises in numerous applications ranging from robot navigation to augmented reality (AR) and virtual reality (VR). Despite recent advances in deep learning, most SLAM systems still rely completely on hand-crafted techniques. In this dissertation, we aim to show how deep learning can benefit SLAM -- in certain cases improving the accuracy, robustness, and even sometimes the efficiency. We show not only that deep learning can benefit SLAM, but in many cases SLAM can actually benefit the deep neural network in return, exposing a mutually-beneficial relationship between the learned and hand-crafted techniques. The dissertation is divided in four thrusts. In the first thrust, a robust visual object SLAM system is presented, which utilizes a custom deep semantic keypoint network to provide the monocular SLAM system with metric scale while allowing it to initialize from a single view. In return, the SLAM system provides valuable priors to the network which allows it to track keypoints on symmetric objects consistently across multiple views, which is typically not possible. In the second thrust, a dense visual-inertial odometry (VIO) system is presented, in which dense geometry is represented as a compact optimizable code that is estimated tightly in the VIO estimator. We show that not only the code estimation can improve the dense geometry's accuracy, but that it can also improve the accuracy of VIO in return. In the third thrust, a monocular depth-aided visual-inertial initialization pipeline is presented. The learned monocular depth is shown to improve the initialization performance in low-excitation scenarios where completely hand-crafted initialization performance is degraded, and in turn the inertial information provides the scale to the monocular depth network which allows it to be used as a compact feature representation. In the fourth and final thrust, an efficient and robust dense visual-inertial SLAM system is presented called AB-VINS. Instead of estimating each feature position separately, as in most SLAM systems, AB-VINS utilizes a monocular depth network to represent the geometry, which is shown to improve the robustness and efficiency. Again, the monocular depth provides useful priors to the SLAM system while the inertial information provides the scale to the network. A new hand-crafted technique for pose graph optimization called the memory tree is introduced along with AB-VINS, which is shown to greatly improve the robustness and efficiency over state-of-the-art methods -- allowing for AB-VINS to solve pose graph SLAM while only relinearizing a constant number of variables no matter the number of keyframes. The dissertation is concluded with possible future research directions.
Electrochemical Deposition of Vanadium and Manganese Oxides for Use as Positive Electrodes In Rechargable Batteries
Zlobin, Alexander
Energy storage technologies represent an essential component in redesigning the electric grid to mitigate CO2 emissions and mitigate climate change. Despite the substantial progress which has been made in recent years in electrification of the transportation industry, battery performance metrics such as the charging rate and time have hindered growth. Charging rate is closely linked to battery stability. For a given battery, charging at a faster rate requires greater overpotential which provides a greater thermodynamic driving force for reaction and promotes out of equilibrium reactions. Therefore, improvements in the stability of battery systems can also have implications for the charging rates which can be applied. The theoretical limits on battery capacity are largely tied to the material composition and crystallographic structure of electrodes, which for the positive electrode (cathodes) are related to the number of intercalation sites versus the weight and volume of the material. However, in practice the availability of these active sites during initial charge and their availability over time is substantially affected by issues in transport, crystal defects and transformations, and the stability of the electrode over time. Therefore, controlling stability and minimizing limitations in transport which hinder kinetics of charge/discharge represent crucial challenges in battery design.
The impact of white androgynous faces on binary gender categorization
Cleary, Samwell Carter
Gender categorization is the use of visually derived semantic information to spontaneously sort individuals into socially established gender groups. Lack of gender-prototypicality disrupts the speed of binary gender categorization (Freeman et al., 2008) and gender non-conformity has been shown to lead to increased gender identity denial (Morgenroth et al., 2023). Limited research has used wholly gender-ambiguous stimuli, however, to examine the categorization of androgynous individuals. In this series of preregistered studies, we investigated the impact of androgyny on the binary gender categorization of faces. Participants categorized masculine, feminine, and androgynous white faces as men or women using a mouse-tracking design. Overall white androgynous faces produced significantly longer response times and less direct mouse trajectories compared to gender typical (masculine and feminine) faces. This indicates a higher level of competition between categories during the categorization of androgynous stimuli as a binary gender category compared to the categorization of gender typical stimuli. These results suggest androgyny acts as a disruptive force on spontaneous binary gender categorization. White androgynous faces were also significantly more likely to be categorized as woman than man, suggesting there are stricter expectations of prototypicality for the category of man compared to the category of woman.
ENGINEERING MULTIMODAL NANOPARTICLES TO COMBAT TRIPLE-NEGATIVE BREAST CANCER
Hoover, Elise Caroline
Triple-negative breast cancer (TNBC) is an aggressive subtype of cancer that accounts for about 15-20% of all breast cancer cases and has a poorer short-term prognosis when compared to other subtypes. Many current chemotherapeutics used to treat TNBC are nonspecific, and, therefore, cause off-target effects and fail to elicit a robust tumor response, which results in higher mortality and recurrence rates. Targeted therapy enabled by nanoparticle (NP)-based drug carriers may prove more effective against TNBC. One promising target for TNBC is the Wnt pathway, which is a known developmental pathway that encourages cell proliferation, migration, and cell fate determination and then transitions to mostly stem cell maintenance in adults. Unfortunately, many cancers, including TNBC, have found ways to upregulate the canonical, β-catenin dependent Wnt pathway in order to promote cell proliferation, migration, chemoresistance, and other oncogenic behaviors. Accordingly, Wnt hyperactivity is strongly correlated with recurrence and metastasis. One method to both target and therapeutically manipulate TNBC cells is by coating NPs with antibodies against Frizzled7 (FZD7) receptors in the Wnt pathway, which are overexpressed in ~70% of TNBC cases but minimally expressed by other cells in the body. When antibody nanoconjugates bind FZD7 receptors, it prevents their activation by extracellular Wnt ligands, leading to downstream inhibition of the Wnt signaling pathway. As TNBC lacks targeted treatment options, the ability to use FZD7-targeted NPs to specifically target and eliminate TNBC cells is particularly appealing. This thesis develops methods to coat NPs with FZD7 antibodies and demonstrates their utility in delivering chemotherapeutic agents or small interfering ribonucleic acid (siRNA) therapeutics to TNBC cells.
The first aim of this thesis focused on developing NPs coated with anti-FZD7 antibodies and/or siRNA targeting β-catenin, a downstream protein in the Wnt pathway that is a key mediator in TNBC’s oncogenesis. The goal of this aim was to investigate optimal parameters for modifying the surface of poly(lactic-co-glycolic acid) (PLGA) NPs with either antibodies, siRNA, or both. The second aim of this dissertation investigated the role of FZD7 targeting on the delivery and efficacy of an encapsulated drug, doxorubicin (DOX). The ability of FZD7-targeted, DOX-loaded NPs to inhibit TNBC cell viability was evaluated in vitro and compared against the effects of DOX delivered freely or in non-targeted NPs. Finally, the third aim of this thesis utilized PLGA NPs to carry both FZD7 antibodies and β-catenin siRNAs to inhibit the Wnt pathway in TNBC cells at both the receptor and effector levels. The RNA interference effects of this platform were assessed through in vitro assays that measured TNBC cell viability, migration, drug resistance, and spheroid formation in response to various treatments.
Collectively, this work demonstrates that antibody-modified polymer NPs can be designed to target FZD7 receptors that are overexpressed on TNBC cells and provide specific delivery of diverse therapeutic cargo including chemotherapeutics and siRNAs. Future research in this field should continue to tailor NP design parameters to achieve optimal efficacy and safety and validate therapeutic potential in in vivo models of TNBC.