2026 Project Descriptions
Social Science Curriculum
In the My Streetscape Summer Research Institute, students are trained as youth community researchers by learning about social sciences research methods, such as surveys, interviews, and ethnography. In teams, students use what they have learned about these methods to design a social science research project to get community input on the engineering projects and technologies they are working on.
Students design data gathering tools, go on field trips where they go into the streetscape to engage with community members and gather data through conversations and observations, and analyze the data they collected before writing their final Research Reports. In the Research Reports, students share what their projects were, how they collected their original data, what they learned from community members, and how the findings informed their engineering project and technologies.
ANALYZING QUALITATIVE DATA WITH A.I. TO IMPROVE THE STREETS OF HARLEM
Faculty Advisor:
Sharon Di
Engineering Mentors:
Yongjie Fu, Baohua Yan, Sihun Kim
Social Science Mentors:
Seon Britton, Jenny Fondren
This summer project engages high school students in the design and implementation of a pedestrian mobility data collection system focused on older pedestrians at urban intersections. The project will guide students through the full pipeline of developing a mobile-friendly app interface to interview elderly pedestrians, record their geographic locations, and collect questionnaire responses related to pedestrian mobility, travel behavior, and perceived safety. The front-end interface will be designed to support intuitive and efficient field data entry, while the back-end infrastructure will be deployed on Google Cloud Platform (GCP) to enable secure data storage, management, and access.
Using the app, students will conduct field interviews with a selected number of pedestrians at an intersection, creating a real-world dataset that combines survey responses with location-based information. After data collection, students will perform preliminary data analysis to identify patterns in pedestrian mobility, accessibility concerns, and safety-related behaviors. Through this project, participants will gain hands-on experience in user interface design, cloud-based backend deployment, field data collection, and data analytics, while also developing a deeper understanding of mobility challenges faced by older adults in urban environments.
Mentees
Manhattan Center for Science and Math
Manhattan Center for Science and Math
Manhattan Center for Science and Math
Loyola School
CRASHMAPPER: TESTING THE VALIDITY OF YOLO A.I. OBJECT DETECTION TO PREVENT STREET COLLISIONS
Faculty Advisor:
Andrew Smyth
Engineering Mentor:
Boshra Khalili
Social Science Mentors:
Seon Britton, Jenny Fondren
This project aims to develop an interactive visual-intelligence application for urban transportation safety analysis using COSMOS cameras. Using traffic-camera video as the primary input, the application will implement an end-to-end workflow consisting of road-user detection, tracking and classification; trajectory extraction and cleaning; trajectory-based safety analysis; and vision-language-model (VLM) screening of selected candidate events. The system will focus on interactions involving pedestrians, bicyclists, motorcycles, passenger vehicles, buses and trucks.
The project is motivated by the need to understand potentially unsafe road-user interactions before they result in reported crashes. NYC public resources, including Motor Vehicle Collisions - Crashes, Vision Zero data, NYC Health traffic-injury resources and traffic-volume counts, provide important information for studying collisions, injuries and exposure. Students will review these resources and investigate the availability and limitations of public near-miss or traffic-conflict data in NYC. This review will establish why video-based candidate near-miss screening can complement crash-based analysis.
The technical pipeline will first identify and track road users and classify them into relevant transportation categories. The extracted trajectories will then be cleaned and transformed into a bird's-eye-view (BEV) representation. Based on the processed trajectories, the application will identify potentially important safety events using interpretable indicators such as minimum distance, converging movements, predicted conflict, sudden stopping and interaction density. A VLM will then screen scenes together with concise trajectory to help explain whether the interaction warrants further human review. Final outputs will include a working application prototype, structured trajectory and safety-event files, BEV visualizations, annotated video clips, and a final report and presentation describing the workflow, findings, limitations and future deployment opportunities.
Mentees
High School for Math Science and Engineering at City College of New York
Columbia Secondary School for Math, Science and Engineering
Manhattan Center for Science and Math
The Clinton School
EVALUATING THE VIABILITY OF SMARTPHONE-BASED GAUSSIAN SPLATTING FOR 3D SIDEWALK MODELING AGAINST A LiDAR BENCHMARK
Faculty Advisor:
Zhengbo Zou
Engineering Mentors:
Christine Suen, Weijia Cai
Social Science Mentors:
Seon Britton, Jenny Fondren
This project aims to develop a low-cost, repeatable, street-level 3D change-mapping workflow for sidewalk-scale urban infrastructure and obstructions. The project will use videos collected from phones or action cameras as the primary input for Gaussian Splatting reconstruction. A compact LiDAR sensor will be used as local ground-truth reference data for validating scale, geometry, and measurement accuracy. GIS tools will be used to translate the 3D reconstruction results into georeferenced maps and planning-ready spatial layers.
The central research question is whether video-derived Gaussian Splatting, combined with dynamic-object removal and LiDAR-based validation, can produce reliable measurements of changing streetscape conditions at the sidewalk scale. This project is motivated by a gap between existing large-scale mapping resources and local urban-management needs: Google Maps, public GIS layers, and city-scale LiDAR provide valuable baseline information, but they are often proprietary, stale, top-down, or not directly designed for repeated local measurement of rapidly changing sidewalk conditions. This project therefore focuses on a deployable workflow for timely, local, street-level 3D documentation.
The main technical contribution is a dynamic object-aware reconstruction pipeline. Outdoor street scenes contain pedestrians, vehicles, vendors, sidewalk sheds, and other occlusions. The project will adapt an existing indoor dynamic-object detection/removal algorithm to outdoor street scenes so that transient moving objects can be removed and more stable static or semi-static sidewalk elements can be measured. The output will include cleaned Gaussian Splatting reconstructions, point clouds, object footprints, sidewalk clear-path measurements, and GIS-compatible map layers. The 125th Street Business Improvement District will serve as an application case for evaluating the workflow in a complex urban environment.
Mentees
High School for Math Science and Engineering at City College of New York
Regis High School
Abraham Joshua Heschel School
Trinity School
EMPOWERING COMMUNITIES THROUGH SMART INFRASTRUCTURE: HUMAN-CENTERED TECH FOR URBAN LIFE
Faculty Advisors:
Jason Hallstrom & Serena Hoermann
Mentors:
Suvosree Chatterjee, Jenifar Selim, Sara Sevilla
Rather than applying technology for its own sake, the six-week program grounded nine student researchers in the local urban environment of West Palm Beach, encouraging them to observe public spaces, attend local government meetings, and investigate real-world challenges surrounding safety, accessibility, civic engagement, and government transparency.
Translating their field observations into actionable tech concepts, the students developed five distinct research projects:
- CitySpeak (Sabrina Kaiser): An AI-based mascot and conversational prototype that uses computer vision to interact with pedestrians, provide municipal info, and summarize citizen feedback to boost civic engagement.
- Digital Transparency Initiative (Levni Apraku-Mensah): An intelligent recommendation system and website prototype designed to help residents search, access, and understand local government records while safeguarding private information.
- SmartWalk (Ryan Calixte and Cassandra Glomann): An AI and computer vision crosswalk system designed to detect pedestrians, analyze crossing intent, and alert drivers to improve intersection safety.
- ProxiRail (Adrian Garcia and Gerald Mano): A retrospective risk-detection system that uses computer vision to track pedestrians and active crossing gates at railroad intersections, quantifying "near-misses" to identify high-risk areas before accidents happen.
- StreetSense (Brian Li, Vanessa Pierre, and Daventzky Edouard): A machine learning sidewalk monitoring system that scans images for structural defects, ranks damage severity, and maps problems onto a dashboard to improve pedestrian mobility and infrastructure reporting.
Mentees
Suncoast Community High School
Palm Beach Central High School
A.D. Henderson University School & FAU High School
Olympic Heights Community High School
Boynton Beach Community High
Suncoast Community High School
A.D. Henderson University School & FAU High School
American Heritage High School
Olympic Heights Community High School