143258 research outputs found
Sort by
Learning object-centric local navigation of quadruped robot from RGB observations
No full textUniversity of Minnesota M.S. thesis. August 2025. Major: Computer Science. Advisor: Karthik Desingh. 1 computer file (PDF); vi, 39 pages.Mobile manipulation requires seamless integration of navigation and manipulation capabilities, yet existing systems struggle with the transition from meter-level navigation to centimeter-level precision needed for manipulation tasks. This thesis tackles the “last-meter problem” by addressing this gap between navigation and manipulation systems.
Existing approaches use 3D model informed pose estimation from depth observations, confining generalization to predefined indoor objects. To remove this prerequisite and enable generalization beyond training datasets, we propose an RGB-only imitation learning framework for mobile manipulator robots such as the Boston Dynamics SPOT.
This thesis proposes an encoder-decoder architecture with cross-attention that processes visual observations from SPOT’s five onboard cameras alongside natural language queries, eliminating the need for depth sensors, object models, and indoor maps. We compare three vision encoders (ResNet18, DINOv2, and Grounded SAM2) and demonstrate that only Grounded SAM2, with explicit object segmentation, successfully generalizes to new object positions (98%) and unseen environments (91%) while maintaining ≤30cm translational and ≤ ±10◦ rotational accuracy. ResNet18 and DINOv2 fail completely, revealing they memorize spatial layouts rather than learning object-centric navigation policies. Category-level generalization remains challenging for Grounded SAM2 (48% success). This thesis work demonstrates that explicit semantic grounding is beneficial for robust object-centric local navigation in real-world environments where precise positioning enables successful manipulation.Mahmudova, Fidan. (2025). Learning object-centric local navigation of quadruped robot from RGB observations. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278719
A dimensional approach to perinatal mania and psychosis
No full textUniversity of Minnesota Ph.D. dissertation. July 2025. Major: Developmental Psychology. Advisors: Megan Gunnar, Robert Krueger. 1 computer file (PDF); viii, 145 pages.Clear and compelling evidence links childbirth with heightened risk for mania and psychosis. Research is primarily limited to perinatal women with categorical diagnoses, despite consensus that symptoms of mania and psychosis are fully continuous in the general population. As with perinatal depression and anxiety, subthreshold symptoms of perinatal mania and psychosis are likely to be empirically and clinically meaningful. Further, these symptoms are likely to be missed with current screening methods. This prospective, longitudinal study examined the prevalence, course, and correlates of dimensionally-assessed symptoms of mania and psychosis in a general sample of 168 individuals giving birth for the first time. Across four timepoints from late pregnancy through 2 months postpartum, symptoms were measured with validated questionnaires. At 2 months postpartum, mother-infant interaction and infant development were characterized with standardized observational protocols. Dried blood spot cytokine (IL-6, IL-10, TNF-α) and hair cortisol levels were determined at three timepoints. As hypothesized, mania and psychosis showed dimensionality and significant longitudinal change. Symptoms of mania symptoms peaked in the immediate postpartum, along with symptoms of depression and anxiety, while symptoms of psychosis declined from late pregnancy to 2 months postpartum. In support of the second hypothesis, higher scores on a mania and psychosis composite were uniquely associated with less optimal maternal sensitivity, after accounting for depression and anxiety composite scores and other key covariates. The mania and psychosis symptom composite also were associated with lower infant cognitive development, though not independent of the depression and anxiety symptom composite. Average levels of perinatal cortisol and cytokines were generally unrelated to symptom dimensions, though in secondary, longitudinal analyses, higher symptoms of mania were associated with higher levels of IL-6, with this finding persisting after accounting for covariates, including depressive symptoms. Findings suggest that subclinical symptoms of mania and psychosis are an important, yet currently underdetected, feature of perinatal mental health in need of further investigation.Howland, Mariann. (2025). A dimensional approach to perinatal mania and psychosis. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278193
2024-2025 UMRA Representative to the UMN Age-Friendly University Annual Report
No full text. (2025). 2024-2025 UMRA Representative to the UMN Age-Friendly University Annual Report. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278697
2024-2025 UMRA Organizational Memory & Continuity Annual Report
No full text. (2025). 2024-2025 UMRA Organizational Memory & Continuity Annual Report. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278582
No full textUniversity of Minnesota Ph.D. dissertation. May 2025. Major: Computer Science. Advisors: Haiyi Zhu, Zhiwei Steven Wu. 1 computer file (PDF); xx, 272 pages.Machine learning-based technologies are increasingly being used to assist care work in high-stakes domains, from provisioning resources for poor families to providing mental health support for people experiencing distress. These technologies have been introduced in hopes that they improve care and decision quality. Depending on how they are designed and used, these technologies may also perpetuate harms like racial discrimination or carcerality. In order to understand how machine learning technologies can harm or help, it is necessary to understand the perspectives of people who use these technologies or are impacted by them. Yet, in many high-stakes domains, the perspectives of impacted people ---especially those who are marginalized or do not have the ability to directly influence the design of these technologies--- remain overlooked. This dissertation presents case studies of evaluating and designing machine learning technologies in child welfare and digital mental health through both quantitative and qualitative methods with people who may be impacted by machine learning technologies. Within child welfare, I explore how existing algorithmic decision-making tools exacerbate harms. First, I evaluate a particular algorithm used in the child welfare system to understand how workers use it to reduce or exacerbate racial biases. Second, I engage impacted people like parents and workers to understand how these algorithmic technologies replicate further systemic harms like carcerality. I then explore how we might design different technologies to benefit those most marginalized by the child welfare system. Within digital mental health, I continue to explore how AI-based technologies might be designed or deployed responsibly in this space, if at all. I use participatory design to understand how digital mental health support providers approach suicide prevention online and whether they think machine learning technologies could benefit them while preventing harms to support seekers. Finally, based on suggestions from mental health support providers, I design and evaluate conversational agents that simulate people in distress to help train new support providers. This work aims to showcase ways to understand how machine learning technologies exacerbate systemic harms and how we might design them better.Stapleton, Logan. (2025). untitled - dbId: 76be8308-49dd-49db-a4a8-d89a71580b38. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278089
November-January-February 2025-26 UMRA Newsletter
No full text. (2025). November-January-February 2025-26 UMRA Newsletter. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278406
The impact of political ideology on consumer behavior
No full textUniversity of Minnesota Ph.D. dissertation. July 2025. Major: Business Administration. Advisor: Akshay Rao. 1 computer file (PDF); v, 128 pages.As the political partisanship divide grows in the U. S., and perhaps world-wide, political ideology has become a prominent individual difference variable that likely impacts people’s decision-making across various domains, including voting preference, daily life choices, as well as consumer behavior, which is the focus of my dissertation. Specifically, I investigate how political ideology drives one’s consumption decisions, which improves lives, strengthens society and might benefit the world at large. This dissertation includes two essays. In Essay 1, I examine the impact of political ideology on one’s preference for self-improvement products. In Essay 2, I investigate the impact of political ideology on one’s preference for volunteering behaviors.Hao, Xianyu. (2025). The impact of political ideology on consumer behavior. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278184
Housewives, writers, and communists: staging domesticity in imperial Germany
No full textUniversity of Minnesota Ph.D. dissertation. May 2025. Major: Theatre Arts. Advisors: Margaret Werry, Sonali Pahwa. 1 computer file (PDF); vi, 297 pages.Domesticity is frequently seen as banal, innocuous, and every day, yet for the German Empire, it was an ideological apparatus rooted in systems of oppression inside and outside of Europe. German women playwrights used their plays to question, critique, and reimagine the labor expected of the German Hausfrau (housewife) at the turn of the 20th century. My dissertation examines domesticity as an aesthetic project and situates the performance of labor as a process of gender embodiment. These playwrights illuminate the oppression sedimented within the body of the Hausfrau and the violence inherent to household commodities imported from global colonies. This project emphasizes the layers of oppression within capitalism: these playwrights unwittingly make visible the colonial lineaments of their own domestic oppression, which intervenes in the existing literature on these subjects by tracing the embeddedness of colonial imports to every facet of German life. My research uses modern feminist and Marxist analysis to examine how theatrical devices and practices uniquely enabled women’s critique of domesticity and how their unmasking of the scripted quality of gender and domesticity revealed the colonial oppression necessary for those systems to function. In doing so, they transformed the living room into a microcosm of the economic, social, and political issues facing Germany, Austria, and the colonization practices of Europe.McKeever, Greta. (2025). Housewives, writers, and communists: staging domesticity in imperial Germany. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278043
Misclassified: California strippers and the politics of rights, refusal and freedom under AB5
No full textUniversity of Minnesota Ph.D. dissertation. 2025. Major: Feminist Studies. Advisors: Tracey Deutsch, Diyah Larasati. 1 computer file (PDF); iii, 183 pages.This dissertation investigates dancers' experiences during the transition to employee status under California’s AB5 to understand why many reject inclusion in a law that is widely celebrated as a victory for workers’ rights. While employee classification is often promoted as a policy solution to the increasing precarity of gig economy labor, dancers’ experiences in becoming employees complicate that narrative. Through social media discourse analysis and qualitative interviews, this study shows that dancers experienced an intensification of labor exploitation and discrimination under employee status and also lost vital strategies to navigate conditions of structural and state violence. Set against the backdrop of the COVID-19 pandemic and the Black Lives Matter uprisings, dancers situate their refusals of employee classification within a broader recognition of the inadequacy of legal rights frameworks to address racialized structural and state violence. Yet, within these refusals, dancers also put forward generative visions for reimagining social cooperation, labor policy, and labor organizing beyond the limits of existing legal and labor paradigms. Freedom becomes a central paradigm through which dancers articulate their goals for their work and lives, suggesting that freedom, rather than inclusion, could serve as an expansive and liberatory foundation for galvanizing a labor movement in an increasingly fragmented post-industrial economy.Turner, Lindsay. (2025). Misclassified: California strippers and the politics of rights, refusal and freedom under AB5. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278100
Episode 192: Ice Cream and Shark Attacks
No full textRuntime 00:56:35In "Ice Cream and Shark Attacks," Dr. Osterholm and Chris Dall discuss the FDA’s recent decision to authorize updated COVID-19 vaccines for limited groups, the confusion around access for those under 65, and the uncertain outcomes of this week’s ACIP meeting. They also cover ongoing upheaval at the CDC, provide updates on COVID-19 and measles, answer an ID Query on Chagas disease, and share the latest installment of This Week in Public Health History. Vaccine Integrity Project Viewpoint: Four tips for understanding this week’s ACIP meeting (https://www.cidrap.umn.edu/vaccine-integrity-project/viewpoint-four-tips-understanding-week-s-acip-meeting); Vaccine Integrity Project - Fall Immunization Information (https://www.cidrap.umn.edu/vaccine-integrity-project/immunization-info).Dall, Chris; Osterholm, Michael. (2025). Episode 192: Ice Cream and Shark Attacks. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/278876