48440 research outputs found
Sort by
Long water waves on graphs
No full textThe physical problem studied accounts for surface water waves in a chan- nel network. The mathematical model refers to partial differential equations (PDEs) on a graph. The goal is to study wave reflection due to compatibil- ity conditions at the vertex of the graph. The PDE system studied is the shallow water (or long wave) model, as considered by Jacovkis (SIAM Appl. Math., 1991). Jacovkis considered converging and diverging Y-shaped junctions, constrained by the Neumann-Kirchhoff vertex conditions. Jacov- kis proved a solvability condition for the linearized hyperbolic system. This dissertation extends his framework to the reflection-transmission of waves at junctions, in different channel configurations, considering balanced and un- balanced graphs. These different regimes are expressed through scattering matrices at the vertices. New numerical simulations, not present in Jacovkis, illustrate the differences between these junction configurations and reveal when a balanced junction is reflectionless, leading to a transparent graph as called in the literature. Examples include novel configurations such as symmetric and asymmetric islands
Development and Validation of a Low-Cost Optical Sensor System for Real-Time Teleoperation of Anthropomorphic Robotic Hands
No full textDexterous robotic hand teleoperation demands data gloves balancing sub-degree measurement accuracy with sub-30-millisecond response latency —capabilities histor- ically limited to commercial systems costing 5, 000to30,000. These prohibitive costs restrict access for research laboratories, educational institutions, and small-scale de- velopers, constraining innovation in hazardous environment teleoperation, rehabilita- tion, and surgical robotics applications. This thesis demonstrates that advanced performance in finger joint angle track- ing can be achieved at consumer-grade cost through infrared optical bend sensing. The developed sensor employs an 850 nm LED and photodiode embedded in silicone rubber tubing, where light attenuation varies exponentially with joint curvature. A transimpedance amplifier with 100 kΩ feedback resistance converts photocurrent to voltage, sampled by a Teensy 4.0 microcontroller at 1 kHz. A simple logarithmic calibration mapping θ = k · ln(Vref/Vout) relates voltage to joint angle, without the need for complex models. Comprehensive validation against three independent measurement systems (i.e. optical sensor, LEAP Hand Dynamixel encoders, and Arducam 4K camera vision tracking) demonstrates strong correlation (Pearson r = 0.9957) across the 0–45◦ operating range. Extended testing over 325 seconds (1,931 synchronized samples) reveals a measurement error of -0.024◦ mean with 1.015◦ standard deviation and 1.08◦ RMSE. Additional characterization quantifies hysteresis (0.96◦ average width) and progressive drift (-0.63◦ over 1,000 cycles), establishing performance boundaries for practical deployment. The complete system costs under 750 compared to 30,000 commercial alternatives, resulting in a 7-to-40-fold cost reduction while maintaining comparable accuracy. This democratization of precision hand tracking technology removes economic barriers to research in surgical robotics, teleoperation of manipulators in hazardous environ- ments, and quantitative rehabilitation therapy. Comprehensive documentation and open-source hardware design facilitate reproducibility and community-driven refine- ment, expanding access to high-fidelity human-robot interaction capabilities previ- ously limited to well-funded laboratories
Modeling Electrospray of Water and Liquid Metals
No full textThis work investigates the feasibility of using electrospray atomization as a novel method for producing powders from refractory metals for use in additive manufacturing. Theoretical estimates of operating voltages and flow rates show that stable electrospray conditions are attainable for liquid metals, suggesting the process is physically viable. Multiphysics simulations in COMSOL were used to model the behavior of water and liquid metals under applied electric fields, demonstrating similar Taylor cone formation and further supporting that electrospray with liquid metals can be conducted successfully under appropriate conditions. A custom water electrospray apparatus was designed and tested to validate the modeling approach, with experimental results aligning with simulations. Together, these results establish a foundational understanding of the electrospray process as it applies to liquid metals and supports its potential use for synthesizing powders from refractory metals for use in additive manufacturing
Osaka Ritsumeikan Japan Project Center MQP 2025 - Shin'ai
No full textThe Shin’ai MQP was developed for IMGD and CS over the span of 6 months. 3 months of pre-production on campus at WPI and the remaining 3 being completed while at Ritsumeikan University’s Osaka-Ibaraki Campus (OIC). Alongside our game, we were also split into different teams alongside Ritsumeikan students to create an additional project using GNSS, a centimeter-accurate location system similar to GPS
The Tourist Experience in Iceland
No full textThe goal of this project was to provide the Icelandic Tourist Board with insight into the experiences and expectations of tourists visiting Iceland. Due to the rise in tourism in recent years, the Tourist Board gathered quantitative data through exit surveys at the international airport to gauge tourists’ interest in Iceland. To supplement their data, we conducted 293 interviews with tourists from 40 countries, asking about their experiences and expectations. We triangulated our data by analyzing 130 online reviews and conducting interviews with several experienced tour guides. Using this data, we identified how these tourists’ expectations were and were not met and why, and we recommended ways the Icelandic Tourist Board might begin to improve the tourist experience in the future
On the Role of Large Language Models for Causal Graph Identification in Economics Literature
No full textLarge language models (LLMs) have shown remarkable performance across diverse domains. However, it is unclear if this impressive performance extends to being an intelligent system. The highest form of reasoning is the ability to causally reason in realistic settings, but there are no benchmarks to assess this. In order to be realistic, it must have varying degrees of confounding, include diverse samples (length of text, size of graph, and domain distribution), and be drawn from the real-world. As published economics papers on causal graphs meet this criteria, we use it as our domain. We introduce RECG-Bench (Realistic Economics Causal Graph construction-Bench), the first benchmark designed to evaluate LLMs in realistic causal graph creation from text in economics. Our evaluations, using an automated LLM-as-a-Judge evaluator, provides a detailed analysis of the strengths and weaknesses of varied state-of-the-art LLMs on real-world causal reasoning. We find LLMs struggle even under the most favorable conditions, regardless of text length, size of graph, or domain, with performance further reduced as degree of confounding increases. Overall, RECG-Bench opens up new avenues for evaluating and analyzing the realistic causal graph construction abilities from text for LLMs
Composting in Aishō Town B25
No full textAishō Town, a rural town with a history of agriculture located in Shiga, Japan, faces low participation in composting programs despite personal gardens being widespread. Our project aimed to understand Aishō and its perception of composting in order to increase participation in its composting programs. Working with our sponsor Ms. Miyoko Kuzutani, the leader of composting efforts in Aishō, we conducted interviews and surveys and made observations at events across Shiga Prefecture and created a public awareness video. We found relatively strong interest in composting, but extremely limited knowledge about how to compost and about existing composting initiatives. We recommend expanding clear and accessible public messaging to help locals engage more easily with Aishō’s composting programs
Developing Accessible Educational Resources for Kaha Garden
No full textTo address limited transportation to Kaha Garden, a xeriscape garden of native Hawaiian plants maintained by Hui o Koʻolaupoko, for students and homeowners who are unable to visit the garden in person, our team created a series of educational resources that includes an ArcGIS StoryMap, a virtual garden walkthrough video, and 3 sets of plant ID cards. We conducted research through websites on native plants to learn about their growth habits and cultural significance to Hawaiʻi, determined what content to include in our resources by holding interviews with volunteers and pedestrians, and piloted our resources with teachers and received feedback to understand how to improve them. These resources will be used by HOK to expand their curriculum and teach more people about native plants