CurbMap 360 was an AI-assisted accessibility-mapping prototype developed from Fall 2024 through Summer 2025. The project explored community sidewalk scanning, quality control, routing support, and anomaly/out-of-distribution screening for unreliable visual scans.
The technical work included autoencoder reconstruction-error screening, VAE latent/OOD analysis, and normalizing-flow density modeling as part of the scan-quality pipeline.
- 1st Place — 2025 Culver Business Model Competition
- Presented in additional entrepreneurship/pitch settings during the project period
The original prototype codebase is not included in this repository. The repository preserves the technical and presentation materials that remain available.
Chaewon Yoon
chaewon.yoon.ds@gmail.com