48440 research outputs found
Sort by
Fostering the adoption of safer A-nets for speed events in ski racing
No full textB-Nets have been a crucial safety feature in alpine ski racing since their introduction in 1985, yet their effectiveness and regulation remain uncertain. This study investigates the effectiveness of B-Nets in preventing injuries and how their design and implementation are influenced by governing bodies. Through an analysis of manufacturer guidelines, governing agency policies, and legal frameworks, we found that B-Nets are largely unregulated, with manufacturers setting their own standards. While organizations like FIS provide recommendations for their use, no enforceable regulations exist, leading to inconsistencies in setup and maintenance. Additionally, limited research and inaccessible injury reports make it difficult to assess the full impact of B-Nets on skier safety. Our findings highlight the need for standardized guidelines, improved oversight, and further research to determine whether B-Nets should play a greater role in ski racing risk management. To begin addressing these needs, we proposed multiple solutions, gathered data from a follow-up survey, and created an axiomatic design for a B-Net absorption experiment
Sustainable Methods for Organic Acid Synthesis
No full textOxalic acid is a versatile organic compound with wide applications in the pharmaceutical, metallurgical, and textile industries. Traditionally synthesized via oxidation of carbohydrate-based feedstocks using concentrated nitric acid, oxalic acid production relies heavily on non-renewable resources and generates significant chemical waste. With growing environmental concerns and increasing pressure to reduce industrial carbon footprints, alternative sustainable production pathways are needed. This thesis explores the feasibility of synthesizing oxalic acid from two major waste streams: polyethylene terephthalate (PET) and carbon dioxide (CO₂). By repurposing these abundant and environmentally persistent materials, this work aims to contribute to the development of a circular manufacturing ecosystem
Computational Complexity Across Models: Communication, Circuits, and Learning
No full textWe are interested in the complexity of computing and representing Boolean functions. Our objective is to provide both upper and lower bounds on the expressiveness of various models of computation, emphasizing connections between circuits, communication, and learning. We establish tight bounds on the communication complexity of deciding whether one string contains another as a subsequence, yielding lower bounds on the sample complexity of an analogous learning problem. We also provide lower bounds for circuits computing the subsequence containment. We study the capacity of the well-known nearest neighbor classification scheme to represent Boolean functions. We show that nearest neighbor representations are closely related to circuits, and that strong lower bounds against this model would constitute a breakthrough in circuit complexity. We also contribute a new dataset of NP-hardness reductions, which we call the Karp dataset. This dataset is designed to be used in conjunction with machine learning techniques to study reasoning capacity of modern artificial intelligence systems, and potentially to discover new NP-hardness reductions
Dynamic Lighting Estimation for Augmented Reality Systems
No full textHigh-quality environment lighting is the foundation of creating immersive user experiences in mobile augmented reality (AR) applications. However, acquiring the environmental lighting is a challenging task due to key limitations in AR devices, as well as environmental and user dynamics. While recent advancements in 3D vision and generative AI have presented new opportunities for estimating environmental lighting, challenges remain in addressing the system latency, privacy concerns, and robustness of lighting estimation systems. These challenges are crucial for dynamic and interactive AR applications, profoundly impacting the real-world user experiences of AR systems. This dissertation addresses these key challenges through four primary research tasks: (1) Mobility-aware lighting estimation. We first introduce LitAR, a framework that integrates spatial variance-aware near- and far-field lighting reconstruction with real-time rendering techniques. By employing guided data capture policies and computational optimizations, LitAR achieves visually coherent rendering with significant improvements in speed and accuracy over existing methods. (2) Privacy-preserving reflection rendering. Rendering reflective AR virtual objects often exposes out-of-camera scene details, posing privacy risks. This dissertation develops ReflAR, a novel system that combines privacy-preserving obfuscation techniques with lighting reconstruction. By identifying and blurring sensitive scene elements, such as human faces and text, ReflAR ensures that high-quality reflective rendering can be achieved without compromising user privacy. (3) Context-guided generative lighting estimation. To address the limitations of device sensing and the robustness of generative models, this work presents CleAR, a robust generative lighting estimation system. Leveraging context-aware diffusion models, CleAR generates high-quality and diverse environment maps while maintaining physical consistency. The system also incorporates real-time refinement capabilities, enabling it to adapt to dynamic lighting conditions and significantly improve rendering accuracy and robustness in mobile AR scenarios. (4) A rethink of edge-assisted perception systems in the era of foundational models. In the final chapter, we extend our research from lighting estimation to generic perception systems on mobile AR. We present FARSight, a prototype edge-assisted and model-based foundational perception system that rethinks the AR perception paradigm
Novel Plateless Tibial Plateau Leveling Osteotomy Implant
No full textThe tibial plateau leveling osteotomy (TPLO) is one of the most commonly performed veterinary orthopedic procedures. As described, the technique utilizes either a procedure-specific compression or locking plate and screws to stabilize the rotated tibial osteotomy. The objective of this study was to assess the ability of a novel plateless TPLO implant to maintain TPLO stability by comparing it through mechanical testing to a locking plate. The novel TPLO implant utilizes a lag screw which locks into an anchored post. The post is designed to distribute the compressive forces uniformly across the osteotomy site through a lag effect of the screw. In addition, the post provides a metallic rim which protects the surrounding bone substrate from localized loading leading to fracture failure
Stock Market Simulation 2503
No full textThis project explores the comparative performance of two investment strategies—Buy-and-Hold and Momentum Trading—through a six-week stock market simulation. Using simulated trades based on real-world market data, the Buy-and-Hold strategy maintained a static portfolio after initial stock purchases, while the Momentum Trading strategy actively bought and sold stocks to capitalize on short-term price movements. Throughout the simulation, weekly reports documented trade history, portfolio value, market conditions, and relevant economic events. The final results showed that the Buy-and-Hold portfolio achieved a total value of 109,695.43, reflecting a higher overall return. The analysis also compared these results with the S&P 500 index performance during the same period. This project highlights key lessons in portfolio management, the impact of market volatility, and the trade-offs between passive and active investment strategies
PACES: Exploring Pool Allocation for Embedded System Heap Security
No full textThis paper explores \textit{PACES}, a conceptual system design to implement memory pooling in embedded systems with the goal of increasing heap security. Embedded systems are commonly programmed in memory unsafe languages like C, which makes them vulnerable to memory-centric attacks through dangling and corrupted pointers. Type-based memory pooling has been shown to mitigate some of the risk of these attacks by ensuring that a given location will only hold one type of object, but have also come with a variety of potential drawbacks. We explore potential design and implementation choices for a pool-based allocation system. We then suggest additional security tools that can be used in conjunction to further increase security benefits. Through this exploration, we suggest the design of a system that ensures that a pointer to a specific object in memory will always retrieve either an object of the correct type or nothing. Additionally, the pointer will always point to the start of an object within the correct type pool and cannot be used to access an object of an incorrect type
An Apparatus for the Electroplating of Chrome-Nickel Carbide Powder with Nickel
No full textChrome-Nickel-Carbide is a thermal spray powder with a high hardness and excellent coating properties on par with traditional chrome coatings. Coating components with this material using cold spray could allow it to be deposited on heat-sensitive substrates and could also enable its use in repair and additive applications, but it's lack of ductility means that it abrades and erodes the substrate at the critical velocity for deposition. To prepare this material for cold spray, we can encapsulate it in a uniform layer of nickel, a more ductile metal, which will deform on impact and allow the Cr-Ni-C to form a coating. This thesis concerns the development and refinement of an apparatus for the encapsulation of Cr-Ni-C particles via electroplating. Experiments were performed over the course of a year, and steps were taken to mitigate micro-scale particle agglomeration, dendrite growth, electrolyte evaporation, and process efficiency. The result was encapsulated Cr-Ni-C with a process efficiency of 73%, a design coating thickness of 1 μm, and minimal agglomeration in the sample. In the future, the application of the apparatus developed here can be extended too different metallic powder applications, including copper on tungsten or copper on chromium
Adapting Gen IV Nuclear Microreactors into the Next Generation Hybrid Reactor for Research and Power
No full textWe propose using Generation IV nuclear microreactors to simultaneously meet the rising demand for carbon-free energy while also serving as next-generation neutron research reactors. We evaluated the deployment of two next generation reactor designs: The eVinci™ microreactor, which is a high temperature heat pipe reactor capable of providing electric power and hot water for building heating, and the MARVEL microreactor, a high temperature liquid metal-cooled test bed reactor. To that end, the energy output of the eVinci reactor was compared to the heat and electric demands of Worcester Polytechnic Institute (WPI) and both reactors were computationally modeled in to determine its research-relevant neutron fluxes and shielding requirements. MARVEL was found to be able to support neutron activation analysis (NAA), prompt-gamma neutron activation analysis (PGNAA) and radiography. eVinci was found able to produce a large amount of useful Mo-99 for imaging while supporting the university research programs across many fields with NAA, PGNAA, neutron scattering, and radiography. Isotope production, research facilities and heating/power are possible concurrently without altering the current design of the eVinci reactor
Computational and Experimental Approaches to Brain Stimulation: TMS Simulation, Coil Measurement, and Neural Structure Analysis
No full textThis MS thesis presents original research on modeling brain stimulation at various levels of complexity and the development of experimental methodologies for transcranial magnetic stimulation (TMS) coils. The work spans computational modeling, automated measurement systems, and the analysis of synaptic orientations. First, a MATLAB-based software tool was developed to simulate the magnetic fields and inductances of TMS coils, providing a flexible framework for designing and analyzing new coil configurations. It allows user-defined centerlines and surface meshes, outputting results in NIfTI format for broader compatibility with electromagnetic modeling tools. Second, an automated measurement setup was designed to facilitate precise mapping of electric (E) and magnetic (B) fields generated by TMS coils. A coordinate measurement machine (CMM) was constructed and integrated with a MATLAB-controlled system, enabling efficient, high-resolution field measurements without manual intervention. Third, a novel algorithm was implemented to extract the centerline (skeleton) from segmented surface meshes of complex geometries. Using iterative Laplacian smoothing and graph-based methods, the approach reconstructs a 1D representation suitable for further electromagnetic analysis. The algorithm also estimates cross-sectional radii and classifies structural elements for spiny neurons, ensuring compatibility with simulation software such as Neuron. Finally, the spatial orientation of conductive structures in the H01 human brain sample was analyzed to investigate potential structural biases affecting electromagnetic field distributions. Future work includes examining the alignment of these structures relative to external electric fields, this study provides insights into optimizing stimulation parameters for improved targeting and efficiency. The findings of this thesis offer tools and insights that can be applied to the design, analysis, and optimization of electromagnetic stimulation for brain stimulation applications