48440 research outputs found
Sort by
Enhancing Physics Education in Underrepresented Communities with a Math-Centered Pedagogical Approach
No full textThis project explored the design and implementation of math and physics lessons within a Worcester Middle School classroom, aiming to address academic disparities through culturally relevant and hands-on instructional strategies. Using reflexive thematic analysis of student feedback, this study found that collaborative, experiential learning approaches enhanced engagement and comprehension compared to traditional lecture methods
AI-Powered 3D Facial Generation for Robotic Interfaces
No full textThis project presents the design and implementation of a robotic interface capable of generating realistic 3D facial expressions and emotions, synchronized with conversational audio generated by a large language model. The system takes a user’s input, produces a text response using the Llama model via the Ollama API, and converts it to speech using pyttsx3. The emotional tone is extracted from the text using a pre-trained DistilRoBERTa model and both audio and corresponding emotion weights are sent to NVIDIA’s Audio2Face engine. Audio2Face then maps these emotion weights, audio input, and facial expressions to a MetaHuman character rendered in Unreal Engine. The result is a generated face displayed on a flexible OLED display attached to a 3D-printed robot head. This work addresses the growing need for emotionally expressive, human-like social robots and demonstrates a low-latency solution to improving human-robot interactions. Our model would be effective in future uses in health care, customer services, assisted care, and education as a social robot
Fixed Wing Micro-Drone Design for the SAE Aero Design Competition
No full textThis project focused on the design of a fixed wing micro-drone for the Society of Automotive Engineers Aero Design West Competition. The team was evaluated through a design report, flight demonstration readiness review, and a flight mission score. An RC aircraft able to carry a water payload with a short takeoff distance to fly a 360-degree circuit was designed. Minimizing the wingspan and empty weight was prioritized in the design to maximize the mission score. The design process included airfoil analysis, wind tunnel testing, stability analysis, electronics and propulsion analysis, and CAD modeling. Flight worthiness was proven prior to competition through glide, taxi, and flight tests. The team placed 8th overall in the micro-class competing with universities from various countries
Vibrational Therapy to Reduce the Effect of Essential Tremor
No full textEssential tremor occurs in approximately 3% of adults 50 to 59 years old and significantly reduces quality of life. Current treatment options fall short, with medications working for about 50% of patients and devices being unaffordable or requiring specific prescriptions from a doctor. Therefore, the goal of our project was to develop an affordable, wearable device to reduce tremors in upper extremity muscles in adults over 55. The team developed a prototype that uses vibration motors to deliver a low frequency, vibrational stimulus targeting the Meissner corpuscles in the fingertips and palm of the hand. To determine its effectiveness, the prototype was tested on individuals with and without essential tremor
CS: MQP: Mapping Global China: Visualization Design and Development
No full textThis report presents a custom web application developed for the Mapping Global China initiative to visualize China's economic and geopolitical influence. The platform, designed for educators, researchers, and policymakers, enables users of all technical backgrounds to explore China’s investments, infrastructure projects, and soft power influence worldwide. Key features include customizable filters, heatmaps, and data layers for analyzing regional and industry trends, along with tools for comparative data analysis. This report details the technical development, design, and processes of the project, and highlights how our team enabled data-driven analysis by improving data usability and accessibility
Shear Stress Sensation of Cancer Cells
No full textMetastatic cancer accounts for over 90% of cancer deaths each year. The phenomenon of metastatic cancer is still being researched, and many mechanisms of its occurrence are unknown. Currently, it is unknown how cancer cells can navigate and survive the circulatory system while maintaining a degree of viability. We ask the question, how are cancer cells able to navigate the circulatory system and what mechanical pathways are responsible for sensing the mechanical shear stress present in the circulatory system and tumor microenvironment? Prior research in our lab demonstrated that TRPA1 is critical for shear stress sensing in the nociceptors of drosophila. We investigate the role TRPA1 plays in the shear stress sensation of human cancer cells by subjecting the cells to shear flow and comparing the shear stress response to cells treated with a TRPA1 antagonist. Our study finds that TRPA1 is required for the shear stress sensation of certain cancer cell lines. With these findings, we suggest that TRPA1 is a potential mechanism for the navigation of the circulatory system for cancer cells during metastasis
Recycling Plastic Waste and its Education
No full textEvery day plastic is disposed of to be processed and recycled, however only 9% is actually recycled globally. This waste ends up in landfills, incinerated, or littered, contributing to negative health effects in humans and broader climate change. The goal of this MQP was to create a direct, in-house recycling facility that could process plastics at WPI to increase the real recycling rate of campus. The recycling process includes plastic sorting, shredding, and extruding into granules or filaments. We designed and fabricated a plastic shredder and extruder to complete the machine processes of this cycle, and tested for recycling PET, HDPE, PP, and PLA. To educate on proper recycling practices, our team developed a website to incentivize students to recycle their plastic waste through a scoring system and a map of campus locations of recycling bins
Evaluating Control Assistance for Latency Compensation in Cloud-Based Racing Games
No full textLatency in cloud game streaming decreases player enjoyment. Various techniques have been devised in order to reduce the effects of network latency. One method is to provide assistance on player controls, which helps the player perform game actions more accurately. Our project implemented and evaluated control assistance for a navigation-based game. We implemented partial AI control over a car’s steering as a form of assistance for players in a top-down 2D racing game, and tested the effects of that assistance with both computer and player-controlled cars. We evaluated our technique through a user study, controlling the amount of latency and the degree of control assistance experienced by players, analyzing the quantitative results to determine the effects on player performance, and the qualitative feedback from questionnaires to determine the effect on player quality of experience. The results showed that latency negatively impacted both performance and quality of experience, while control assistance had a positive impact on both metrics
Integrative Analysis of Large Genomic Data
No full textAs the most common form of genetic variation in the human genome, single nucleotide variants (SNVs) are widely used as primary biomarkers in investigating genetic etiology. Statistical association tests are commonly applied to assess the relationship between complex traits and genetic variants. However, single-variant analyses often have low power, especially for rare variants. To address this, variant-set analyses that group multiple variants have been developed to enhance statistical power. Additionally, incorporating external biological information can prioritize potential causal variants and further improve the detection of genetic risk factors. Existing integrative association tests commonly utilize weighting schemes to combine the biological characteristics of SNVs. However, identifying the best weighting schemes to fully leverage relevant information remains an open question. Moreover, a significant proportion of the heritability of many complex traits is still missing, emphasizing the need for more powerful methods. This dissertation aims to fill these gaps and develop powerful integrative association tests with optimal weighting schemes. First, we study a fundamental problem for designing powerful variant-set association tests: which genetic association approach provides the best signal strength for the variant set? We systematically compare three commonly used approaches in SNV-set analysis: the marginal model fitting approach, the joint model fitting approach, and the decorrelation approach. We demonstrate that the marginal model fitting generally achieves a higher signal-to-noise ratio and validate this finding by comparing summation-based and supremum-based SNV-set tests. Our extensive statistical simulations and a real-data application confirm the advantages of the marginal model fitting approach, highlighting its benefit in designing powerful variant-set association tests. Second, we provide an optimal weighting scheme by optimizing asymptotic efficiency metrics for a general family of tests. We conduct systematic simulations to demonstrate the enhanced power of our approach compared to existing methods. We also apply our method to a genome-wide gene-based association analysis of osteoporosis. The results illustrate that our optimal weighting scheme effectively integrates useful biological information and boosts the power of integrative association tests in real-world applications. Third, we propose a novel framework for integrative association tests with optimal weights in whole-genome sequencing studies, along with a computational tool to facilitate its application. This framework, referred to as the GLOW (inteGrative anaLysis using Optimal Weights) procedure, is flexible to both individual-level and summary statistics data. We apply the proposed framework to a gene-based analysis using summary statistics from an amyotrophic lateral sclerosis (ALS) meta-analysis. Our framework successfully identified well-established ALS risk genes and several putative candidates associated with ALS-related traits and biological pathways, highlighting its potential for discovering novel genetic risk factors for complex traits
Barren, Irregular, Chaotic Terrain Ring Model For Lunar Wireless Applications
No full textThere has been a renewed interest for additional lunar missions, many of which consider deploying a permanent lunar base. To conduct proper link budget analysis or wireless infrastructure planning, an accurate wireless channel model for the barren, irregular, and chaotic lunar terrain needs to be developed. Barren Irregular, Chaotic Terrain Ring Model (BICTR) is a new approach conceptually based on Jake’s model and its ring of reflectors. BICTR is purpose built for the lunar environment and uses the geometry and regolith properties of the irregular terrain. Several deterministic and stochastic processes were integrated to provide an accurate multipath model wireless channel model. These processes include, but is not limited to, Free-Space Path Loss (FSPL), an iterative reflector search process, randomized phase shifts, and Rayleigh Fading. Comparisons with other models such as the Irregular Terrain Model (ITM) and the Irregular Terrain with Obstructions Model (ITWOM) with the NASA Desert Research and Technologies Studies (NASA DRATS) analog measurement campaign as the ground truth, suggest that BICTR is consistently more accurate when predicting signal strength and coverage