48440 research outputs found
Sort by
Quantifying Residential Sources of PFAS
No full textPer- and polyfluoroalkyl substances (PFAS) are persistent synthetic chemicals widely used in consumer products—such as nonstick cookware, water-resistant fabrics, and personal care products—leading to their presence in residential wastewater. This study analyzed PFAS accumulation in the scum, supernatant, and sludge layers of septic tanks from three homes and explored connections between household product use and PFAS contamination. Samples were collected and tested using LCMS, and residents were surveyed regarding their household product usage. Results showed that PFAS primarily accumulated in the scum and sludge layers, with variations linked to specific household behaviors. These findings offer insight into PFAS partitioning and potential mitigation strategies for septic systems
ARFlow+: A High-Performance Framework for Precise Multi-Device AR Experimentation
No full textConducting fair evaluations of mobile AR systems is often difficult due to the complexities in experiment setups. For instance, collecting precise data and ensuring synchronization across multiple devices in real-world, dynamic environments can be highly technically challenging. Traditional frameworks, primarily designed for single-device operation, lack robust multi-device support and struggle to maintain tight temporal and spatial alignment. Such limitations undermine the reproducibility and scalability of AR experiments. In this work, we introduce ARFlow+, a high-performance data-streaming toolkit for AR experimentations. ARFlow+ addresses limitations in existing systems and provides multi-session and multi-device AR data streaming support with both spatial and temporal synchronization mechanisms. Specifically, ARFlow+ achieves accurate spatial synchronization by using a high-precision spatial registration protocol implemented with ArUco markers. For temporal synchronization, ARFlow+ uses a robust Network Time Protocol (NTP)-based mechanism that aligns device clocks with microsecond-level precision. Furthermore, ARFlow+ employs efficient batching strategies across its data collection and processing pipeline, significantly enhancing overall performance. We evaluate ARFlow+ with both synthetic and simulation-based testbed setups. Our evaluation shows that using ArUco markers for spatial alignment results in approximately 4° of rotational offset and negligible translation error, ensuring accurate cross-device registration. Additionally, our NTP-based temporal synchronization can reduce the multi-device data stream time misalignments from approximately 720ms to 72ms, a 90% reduction compared to baseline methods that use native system clocks. Finally, the data batching design significantly improved system throughput, enabling up to 5 simultaneous devices to stream 30 FPS color data—a capability not supported by its predecessor. GitHub repository: https://github.com/cake-lab/ARFlow Video demo: https://youtu.be/Da3Ao2f6UUo?si=qIvUS86eI3So-7j
Electrical and Mechanical Redesign of a Humanoid Robot to Assist Locomotion
No full textAva and Finley are two 3D-printed humanoid robots that are designed for at-home assisted care living for elderly patients. This project builds upon previous iterations of an open-source project by upgrading their electrical and mechanical systems to enable unassisted standing and assisted walking. These developments focused on reducing material and visible wiring, optimizing internal space, and increasing power economy. Mechanical redesigns, informed by force and torque calculations, improved durability and reduced motor slack which is an important parameter for biped stability. Impact testing validated system resilience while material testing and structural reinforcements informed by finite element analysis (FEA) simulations improved overall strength. The electrical system overhaul included multiple iterations of power and communication distribution which optimized performance under varying conditions. These combined modifications in electrical and mechanical efforts show how iterative enhancements can significantly improve performance, longevity, operating efficiency, and reliability which is required to support the robot during locomotion
Sustainable Energy Awareness: Communicating the Benefits of Electrification in New Zealand
No full textNew Zealand’s shift to clean energy relies on widespread household electrification. This project supported this transition to electric appliances by creating a website to inform and motivate the public about the benefits and steps towards electrification. The team interviewed experts and community members, conducted surveys, and analyzed five international websites to identify key motivators and effective design strategies. Cost savings, environmental impact, visual clarity, and step-by-step guidance emerged as top priorities for effective communication. The team then developed the Aotearoa Electrification Hub, featuring comparisons, clear visuals, and practical guides. A user study yielded positive feedback on design and usability and identified gains in user knowledge and motivation
Sustainable Energy Microgrid Alternative for Gateway Park
No full textMicrogrids offer a sustainable solution to the global challenge of climate change from greenhouse gas emissions. To advance WPI’s sustainability goals, this project evaluated the potential for a microgrid at WPI’s Gateway Park through case studies of university microgrids and interviews with campus energy and sustainability experts. Focusing on sustainability, financial, educational, and energy resource aspects, we identified key successes, challenges, and considerations of microgrid systems on university campuses. Findings include ensuring a strong return on investment (ROI), establishing the microgrid as an educational tool, and other informed recommendations for developing a microgrid at WPI
Rural Perceptions on Water Security and Disturbances in Mandi District, Himachal Pradesh
No full textThis study investigates rural perceptions of water security and environmental disturbances in Himachal Pradesh, focusing on villages in the Mandi district. Through a mixed-methods approach, combining structured surveys and interviews, we identified village-level disparities in water access, infrastructure, and institutional responsiveness. Key contributors to water insecurity include landslides, erratic rainfall, and fragile pipe systems. Our findings highlight the need for upgrading water infrastructure, improving road access to isolated communities, and better funding the Jal Shakti Vibhag to ensure timely responses. Community-led solutions like springshed management and rainwater harvesting remain vital for long-term resilience
Blueprint for Demokratihus: Renovation for a Stronger Youth Democracy
No full textUngdommens Demokratihus (UDH) has the potential to become an active hub for youth-led political activities in Copenhagen but is limited by their outreach efforts and volunteer network. We sought to develop stronger methods of outreach and a more structured volunteer network. Through observations, surveys and interviews with volunteers, community organizations, and users of UDH, we identified key aspects of the current volunteer network, organizations of interest for UDH collaboration, and feedback from current users of the space. Additionally, we analyzed how organizations used the space. To address areas for improvement, we created marketing initiatives, a guidebook for volunteers, and met with volunteers to describe our recommendations
Determination of a Standard Set of ‘Persistence Factors’ for Behavioural Change Measures in UK Industrial, Commercial and Public Sector Buildings
No full textThe effects of sustainable behavior changes are an unsaturated area of study in terms of reducing emissions in businesses. The goal of this project was to assist the Midlands Net Zero Hub in researching sustainable behavior changes, their effectiveness, and what implementation methods are best to encourage employee adaptation. Research was conducted by interviewing experts in the field of sustainable behavior change and reviewing case studies. Through these methods, the research team was able to identify key implementation techniques and behaviors that were most effective for reducing carbon emissions. The research team found that financial gain, leadership, education, and energy mapping were the most important themes when motivating businesses to transition to more sustainable behaviors
Responsible AI Adoption Among Non-Profits in Germany
No full textSince artificial intelligence (AI)’s widespread introduction to industries around the world, it has had an undeniable impact on almost all facets of life, particularly in the workplace. To keep up with changing industry standards, businesses and organizations are rapidly exploring how to adopt AI. However, AI raises complex ethical and practical questions. Nonprofit organizations illustrate the need to address these questions carefully, given their relatively limited resources and their strong ethical commitments. Although larger organizations may disregard the ethical concerns of AI implementation in the pursuit of rapid growth and advancement, non-profit organizations must strike a careful balance between the ethical and practical impacts of adopting AI. The goal of our project is to help a small non-profit organization called Service Learning in Deutschland (SLIDE) explore how they can responsibly adopt AI into their workplace. SLIDE hosts seminars and trainings for teachers and administrators across Germany, all to further the influence of service-learning. SLIDE places high value on privacy, protecting the environment, and educating the German youth. Our objective was to help SLIDE understand how to use AI to enhance their workflow safely and ethically. First, we conducted secondary research about AI use in other non-profits and the ethical concerns involved with AI, as well as interviewed local subject matter experts. In Germany, we interviewed SLIDE’s employees to learn more about their organization’s structure and the ways in which we could help them improve their workflow. While conducting interviews, we researched AI applications, specific tools to fill SLIDE’s needs, and the ethical concerns surrounding the tools’ implementation. With this base of knowledge, we recommended the best tools for their needs: an AI transcription tool, Generative AI (GAI) chat bots, and a customer relations manager (CRM). To help SLIDE and other organizations develop ethical approaches to using AI in their workplace, we also developed two written deliverables: a set of “Employee AI Use Guidelines” that was written specifically for SLIDE, and an “AI Ethics Discussion Guide” that can be used by SLIDE and other small non-profits. For other small non-profits, this guide introduces the general use cases for AI, explains the associated ethical concerns, and prompts critical thinking regarding these topics
Learning Warmstart: Accelerating Trajectory Optimization with Self-Attention Siamese Network
No full textThe computational complexity of trajectory optimization poses significant challenges for real-time robotics applications, where timely solutions are critical. This thesis introduces a novel approach to accelerate trajectory optimization by learning to warmstart the optimization process using a Self-Attention Siamese Network. The proposed method maintains a library of previously solved trajectory optimization problems and their solutions, then employs a neural network to identify the most similar past problem when faced with a new optimization task. The solution to this similar problem serves as an initial guess for the new problem, potentially reducing solve time. The approach is evaluated on two dynamic systems: an inverted pendulum and a quadrotor. Results demonstrate that appropriate warmstarts can significantly improve optimization outcomes, particularly for torque-constrained problems. For quadrotor trajectory optimization, the Self-Attention Siamese Network consistently outperforms both random selection and Euclidean distance-based methods, achieving up to 20\% reduction in solve times. The thesis also highlights the asymmetric relationship in warmstarting, where trajectories with lower torque constraints provide excellent initialization for problems with higher torque limits, while the reverse significantly degrades performance. These findings emphasize the importance of constraint-awareness in trajectory libraries and demonstrate the potential of neural network approaches to accelerate optimization for robotic systems