33760 research outputs found
Sort by
DEVELOPMENT AND DEPLOYMENT OF A SOFT ROBOTIC GRIPPER: HARNESSING PRESSURE FOR OBJECT RECOGNITION THROUGH DEEP LEARNING
No full textSoft robotic manipulators are uniquely suited to handle objects with diverse textures, shapes, and stiffnesses, thanks to their highly deformable structure; however, this same characteristic
complicates the collection and interpretation of data for exteroceptive sensing. This challenge
is often overcome by using machine learning on a wide array of strain sensors, but this work
introduces a methodology that leverages the time-series pressure response of a novel, buckling-enabled soft robotic gripper to enable passive sensing capabilities. The behaviors of this pneumatically
controlled, thin-shelled gripper are investigated with numerical modeling and validated experimentally.
We detail a robust manufacturing strategy designed to minimize defects and demonstrate several
representative use cases of the gripper. To develop built-in sensing capabilities, a dataset of the
time series pressure responses during a variety of grasping events is constructed and analyzed.
The dataset is initially explored for qualitative insights into exteroceptive sensing. It is then
used to train two machine learning classifiers to distinguish between the geometry and size of a
given object at over 85% accuracy across multiple classes. Finally, the ability to recalibrate the
model on a new gripper using transfer learning is demonstrated via two grippers with intentionally
produced dimensional modifications. By eliminating the need for embedded electronics or structural
changes, this strategy enables rapid, low-cost haptic feedback from a single interaction, with
broad implications for constrained environments such as surgery and industrial automation
Phenotypic heterogeneity of stem cell-like cancer cells isolated with microfluidic-enabled one-cell culture
No full textCancer metastasis is a complex process and remains the leading cause of cancer-related mortality. This is posited to be driven by a subpopulation of stem cell-like cancer cells, which possess self-renewal potential, differentiation capacity, and resistance to conventional therapies. Effective isolation and culture of these cells are the foremost steps for understanding their biology and developing targeted treatments. However, conventional methods, such as surface marker-based isolation and suspension culture, are limited due to cancer cell heterogeneity and the difficulty in maintaining stemness. To overcome these issues, we employ a bioinspired one-cell culture approach using hyaluronic acid (HA)-enriched alginate core-shell microcapsules to create a microenvironment that selectively promotes the survival and proliferation of stem cell-like cancer cells while inducing cell death in non-stem cell-like cancer cells. Furthermore, despite significant progress in cancer research, stem cell-like cancer cells in metastatic lesions and circulating tumor cell (CTC) populations remain underexplored. These cells may differ from stem cell-like cancer cells within primary tumors. In this study, we extend the application of the one-cell culture to patient-derived CTCs and cells from metastatic lesions, allowing for a more clinically relevant investigation of these stem cell-like cells.
To identify and characterize these distinct subpopulations from both CTC populations and metastatic sites, we characterize multiple phenotypic properties. First, we assess cellular morphology and cancer stemness in terms of self-renewal ability, and protein expression of cells isolated using the one-cell culture method. In addition, we evaluate the mechanophenotypes of stem cell-like CTCs using real-time deformability cytometry and explore the role of YAP (Yes-associated protein), as it is a key regulator that controls the mechanical phenotypes in stem cell-like behavior. Lastly, we examine the dissemination behaviors of the stem cell-like CTCs using intravital imaging in a zebrafish xenograft model, which offers a dynamic view of metastatic potential in vivo.
Together, this dissertation presents a comprehensive phenotypic analysis of stem cell-like cancer cells isolated using the one-cell culture method across diverse cancer types and patient samples. The phenotypic features, including cellular morphology, self-renewal capacity, protein expression patterns, mechanophenotypes, and metastatic dissemination behavior in zebrafish may contribute to the development of therapeutic strategies targeting stem cell-like cancer cells
METRIC GEOMETRY OF FINITE ENERGY CLASSES IN BIG COHOMOLOGY
No full textThis thesis investigates the metric geometry of finite energy classes in big cohomology. These finite energy classes are made of functions that correspond to singular metrics on compact K\"ahler manifolds. These spaces of functions were introduced to find the canonical K\"ahler metrics. We extend their study to big cohomology classes.
On the space of finite energy potentials where represents a big cohomology class, we construct a complete geodesic metric . We show that several metric properties of are the same as in the K\"ahler setting.
In the end, we study the space of geodesic rays in , , and construct a chordal metric on it. We show that is a complete geodesic metric as well
CHARACTERIZING STRUCTURAL COMPLEXITY OF THE EARTH’S FORESTS WITH GEDI
No full textThis dissertation helps advancing the characterization, mapping, and monitoring of forest structural complexity through spaceborne remote sensing centered on lidar technology. It develops three interconnected methodologies that collectively contribute to our ability to assess forest structure across spatial scales. First, I introduce the Waveform Structural Complexity Index (WSCI), derived from the relationship between airborne laser scanning measurements and GEDI spaceborne lidar waveforms, enabling global mapping of actual forest structural complexity, which reveals distinct biome-specific patterns, with tropical forests exhibiting consistently higher canopy complexity than temperate forests. Second, I address GEDI's sampling limitations by developing a computationally efficient deep learning framework that fuses GEDI WSCI estimates with synthetic aperture radar data to produce continuous high-resolution (25m) maps of structural complexity across global forests. This approach demonstrates robust performance across diverse biomes and time periods, while providing calibrated estimates of structural complexity and pixel-level uncertainty. Finally, I apply this fusion dataset to examine forest responses to fire disturbances in a protected area in the Amazon, revealing that fire effects extend up to 2200m into undisturbed forests, with structural complexity showing higher sensitivity to edge effects from fire disturbances than canopy height. These findings collectively contribute to our understanding of forest structural complexity across scales and establish new methodologies for monitoring forest ecosystems in an era of rapid environmental change, while providing tools for supporting forest conservation, management, and research globally
Avoiding Singularities and Spurious Maximizers in the Maximum Likelihood Estimation of Gaussian Mixture Models: Relative Class Constraints Enabled by Metaheuristic Optimization
No full textThis research addresses a long-standing challenge in the maximum likelihood estimation of finite mixture models: the emergence of singularities and spurious local maxima due to an unbounded likelihood function. Traditional optimization methods, such as the Expectation-Maximization (EM) algorithm, are highly sensitive to initial values and often converge to degenerate solutions. While constraints on the parameter space are essential to prevent convergence to these degeneracies, the EM algorithm cannot accommodate what the literature suggests are the most effective type, relative constraints across classes. To address this shortcoming and overcome these limitations, Simulated Annealing with Predicted Constraints (SAPC) is introduced as an optimization framework that incorporates relative class constraints derived from class separation metrics, including the Jensen-Shannon (JS) distance. By integrating these constraints, SAPC confines the search to a compact region of the parameter space that is more likely to contain the true model parameters. These constraints, which are predicted directly from the input data, also extend naturally to tests of class enumeration. Specifically, if the predicted class separation intervals for any pair of classes contain the 95\% cutoff expected for data generated from a single class (empirically determined to be 0.084 for 2D data), the hypothesis that these two classes represent a single class cannot be rejected and, consequently, the hypothesis that the data contain G-1 classes cannot be rejected.
Simulation studies compare the SAPC method to the EM algorithm, evaluating class enumeration accuracy and parameter bias. Results indicate that SAPC significantly reduces parameter bias in the low class separation conditions
Studying the Color, Brightness, and Outbursts of Comet 12P/Pons-Brooks
No full textComets are large objects primarily made up of dust and gas that orbit around the Sun. Comet 12P/Pons-Brooks is known for having massive outbursts, which is when a comet becomes drastically brighter. The outbursts can be unpredictable in cause and duration. There are many theories as to why these outbursts happen, but we don’t always know the causes. Through this semester, we studied the mass of the outbursts and the lightcurve of this comet to learn information about the causes of these outbursts.
We started with plotting the magnitude of the comet versus the distance of the comet from the Sun. We did this to calculate the scale of the outbursts and other information as well. We found that the data, which was in three different filters, didn't line up. We created a loop using python code that found the offset values in each filter and corrected them. With this information we can learn about the scale of these outbursts as well as do some calculations to figure out how massive these outbursts are.
Our next few steps involved a multitude of calculations to find the mass of an outburst on any given day. Through aperture photometry, an astronomical imaging technique, we found the brightness of outbursts on nights of outbursts. We used these numbers to calculate the differential size distribution of the comet, the number of dust grains in the comet’s outbursts, the cross sectional area of the comet, and finally to calculate the mass of the outbursts. The mass of the outburst on July 21st 2023 was found to be 5.18*109 kg. We also compared this qualitatively to other masses to get an idea for just how big this was
Affordable Real-Time CV for Disaster Relief and Beyond
No full textGemstone Team ANDRRSearch and rescue operations following natural disasters are critical for saving the lives of individuals trapped and in need of aid. However, many of the technologies used in search and rescue (helicopters, planes, boats, etc.) can be prohibitively expensive for nations with low GDP. This issue is only exacerbated by the higher number of deaths due to natural disasters in such nations due to less resilient infrastructure. Unmanned Aerial Vehicles (UAVs) have the potential to replace more expensive technologies in the search for endangered individuals and analysis of at-risk hardware without endangering rescue personnel. While advancements in low-cost UAVs for commercial and hobby use in addition to the development of lightweight computer vision (CV) software and dedicated processors have set
the stage for low-cost search and rescue UAVs, stand-alone UAV costs and part scarcity still pose challenges. Team ANDRR developed a real-time CV module for UAVs to aid in natural disaster relief using a wide range of compatible commercially available parts wherever possible. By conglomerating existing systems into a single effective and easy-to-use unit, we were able to identify individuals in real-time at low cost. The development guide and software for the module are provided to provide increased access to real-time CV processing for search and rescue and other operations
Violence, Stigma, and Discrimination against Trans People in Maryland
No full textThe Maryland Trans Survey is a community-based research project conducted by Trans Maryland and the Queer/Trans Collective for Research on Equity and Wellness examining experiences of trans people in the State of Maryland in areas such as health and healthcare, employment and economic well-being, and legal and policy experiences. To date, this is the largest survey of trans people in the State, with 750 trans people representing all 23 counties in Maryland and Baltimore City.
Data were collected from June to December 2023 through in-person and online community outreach. The project was approved by Towson University’s Institutional Review Board (Protocol #1897) and used Transgender Research Informed Consent (TRICON) Disclosures to provide trans community members with additional transparency on the project, recognizing long histories of harmful practices in trans research from scientific institutions.
Trans and nonbinary people are significantly more likely to experience violence than cisgender people. Violence takes many forms (i.e., physical or sexual assault) and comes from different sources (i.e., perpetrators can be strangers, friends or family, law enforcement, etc.). Violence and discrimination against trans people is a form of gender-based violence, as power imbalances associated with their gender identity make this community more vulnerable to harm and less able to access resources in the wake of violence. Perpetrators frequently leverage anti-trans bias to exert power and control over trans people. Additional prejudices, including sexism, racism, and ableism intersect in unique ways that may increase a person’s risk for experiencing violence. Trans individuals who experience violence are more likely to suffer from negative physical and mental health outcomes.
This brief contains information from the survey related to experiences of trans people with violence, stigma, and discrimination to help advocates, policymakers, and community-serving entities better understand and support the current needs of trans people in Maryland.University System of Maryland - Wilson H. Elkins Professorship (2021-2023); Washington University in St. Louis - Audre Lorde Distinguished Professorship (2023-present); University of Maryland, College Park – Department of Behavioral and Community Healthhttps://transmaryland.org
Supporting the Knowledge Needs of Community Wealth Building Initiatives with a Scoping Review and Subject Guide
No full textThis is a poster presented at Social Equity Leadership Conference (SELC) 2025
Nutrient Uptake Shifts During T4 Phage Replication in Escherichia coli
No full textThe rising threat of antibiotic resistance has prompted renewed interest in alternative therapies, such as phage therapy. Phage therapy offers a targeted approach by lysing specific bacterial hosts, potentially circumventing antibiotic resistance. Our study investigates the host-pathogen interaction between Escherichia coli and T4 bacteriophage, a virus that targets only E. coli, with a focus on metabolic changes during infection. E. coli relies on various carbon sources to fuel its metabolic pathways, and viral infection can reprogram host metabolism to prioritize viral replication, often depleting cellular energy and nucleotide pools. Previous studies on other viruses have shown variable dependence on nutrients such as glucose and glutamine for replication. To determine nutrient utilization by T4 phage, uptake assays adapted for bacterial cells were conducted to examine glucose and glutamine consumption in uninfected versus phage-infected E. coli cells. Infected cells showed a marked increase in uptake of both glucose and glutamine, suggesting that both are critical for efficient viral replication. Ongoing research using glutaminase and glutamine synthetase knockout strains will help clarify the specific roles of these pathways in supporting phage propagation.FIRE: First-Year Innovation and Research Experienc