48440 research outputs found
Sort by
AI Literacy Resource
No full textTo a non-expert member of the public, modern artificial intelligence can seem like a black box. This project aimed to combat that by creating a resource for improving one’s AI literacy. It accomplishes this through: text based explanations, interactive technical visualizations, and links to additional resources. While its main focus is on generative AI technologies, most members of the public only interact with generative AI and they serve as a useful starting point for understanding other kinds of AI
AI-powered Recruitment for International Student Programs at Chinese Universities
No full textAs Chinese universities expand their global reach, and as artificial intelligence (AI) technology advances, understanding how AI can enhance international student recruitment has become increasingly important. This project assesses applying AI-powered recruitment tools to international student programs at Chinese universities. This study investigates how similar technologies can enhance higher education admission. Through interviews, literature reviews, and surveys, we identify pain points such as language barriers, limited information, and lack of data. Our research also explores desired AI features such as multilingual autoreplies, automated content generation, and visualized recruitment data to assist with future product development and implementation
Optimizing Biki for First/Last Mile Infrastructure
No full textWe present this project to assist Biki, operated by Secure Bike Share, close first/last mile gaps in Honolulu's bikeshare program. Using community surveying, GIS mapping, and comprehensive data analysis, we created plans to improve the connection between Honolulu's bus system and the bike share systems. We also provide Biki with a comprehensive improvement plan highlighting low-accessibility areas. This plan includes the introduction of way-finding signage to bus stations, expansion plans supported by GIS suitability maps, and data-driven analysis methods for quantifying the success of their bike share network. Through these measures, we aim to help Biki expand its ridership and to incentivize the use of bike sharing as a healthier and greener alternative to car travel in Honolulu
Kinetic Severity Factor Analysis of HTL and STL of Biomass
No full textHydrothermal liquefaction (HTL) and solvothermal liquefaction (STL), HTL with a co-solvent i.e. methanol, reduces biomass to bio-crude oil which can be upgraded to produce petroleum-like products and fuels. HTL processes are run based on parameters like reaction time and temperature. These variables can be uniformly defined using the kinetic severity factor model (KSF). This study addresses the effects kinetic severity factor on HTL and STL with switchgrass and pinewood. Several conditions were tested and products were calculated on a mass and organic basis for comparison. KSF was observed to show a possible optimum value for oil yield in pinewood STL with clear trend for both feedstocks. Char showed a minimum KSF value to drastically reduce STL run yields, and a consistent trend for HTL
The Evolution of Venetian Cartography
No full textThe preservation of heritage is a challenge that every historic city and culture faces, and cartography is a very valuable tool in addressing this. SerenDPT, our sponsor, realized the importance that improving the collection of cartographic history has in preserving the heritage of Venice. Our project aimed to improve the documentation and application of cartographic tools that SerenDPT employs through digitizing its map collection and researching ways to improve the online experience for viewers. We digitized a total of 385 maps from SerenDPT’s and Roberto Ferruzzi’s collections and proposed improvements to SerenDPT’s already existing tools and proposed new tools. This work will improve the preservation of Venetian culture as a whole
A Handbook for Robotics
No full textThis deliverable presents a beginner friendly robotics handbook for beginner to intermediate students focused on turning ideas into working programs. The book covers common robotics topics in a concrete, digestible way, emphasizing clear reasoning and practical implementation. In several chapters, the material is developed through a problem-first workflow: we start from a simple scenario, make an honest attempt, identify what breaks, and refine the approach into something more reliable. Throughout, examples are chosen to be easy to visualize so readers can focus on building transferable problem-solving habits. As an additional resource, we provide a lightweight simulation environment and example scenarios for hands-on experimentation and iteration before moving to hardware
Improving Practitioner eModules for Youth in Transition
No full textThe goal of this project was to create engaging professional development videos for practitioners who help young people transition from Out of Home Care and a user guide on the video creation process for the Brotherhood of St. Laurence’s Better Futures team. Research for this project encompasses young people in transition from Out of Home Care, instructor training, engagement in learning, educational video design, and interview methods. We incorporated both feedback regarding online learning preferences gathered in practitioner interviews and research from literature into professional development videos. We then produced a user guide that details how to effectively create future videos and recommended that gamification and further interactivity be implemented
Promoting Sanitation Equity in ECD Centres in South Africa
No full textOur project, sponsored by the Centre for Early Childhood Development (CECD), aimed to spread awareness about and promote sanitation equity throughout Early Childhood Development (ECD) centres in South Africa. Using documentation, interviews, and children’s activities, we found major gaps in sanitation among ECD centres and investigated how children and teachers feel about their situations. We shared this information with the public in the form of a short documentary, social media campaign, and museum exhibit
Clustering NBA Seasons Using TDA
No full textThis project applies Topological Data Analysis (TDA) and k-medoids clustering to explore structural patterns in league-wide NBA team statistics across 51 seasons. The persistence diagrams for each season are computed and converted into persistence images which capture topological features of the point clouds. The persistence images are then grouped into five distinct clusters that correspond to meaningful eras in league evolution using the k-medoid clustering technique. These clusters are given as (1) modern peak-optimization, (2) experimental pre-modern, (3) old school, (4) transition to modern playstyle, and (5) physical and defensively dominated
Ava and Finley: 3D Printed Humanoid Robots - Recruiting ECE Students Only
No full textThis project encompasses investigations into computer vision tools for the YOLO model framework built to expand the functionality of the Finley humanoid robot, and help streamline future development. These tools include a training pipeline utilizing WPI's Turing Research Cluster, a general purpose detection utility capable of desired object filtering, and application specific tools. Application specific tools included an oximeter digit extraction tool and chess board recognition pipeline capable of converting images of chess boards to digitized FEN notation