List All Projects List Graduate Projects List Senior Design Projects List REU Projects Other Undergraduate Projects
Project # 14
Development of a Vision-Based Smart Cane with RGB-Only Perception for Collision Avoidance and Scene Recognition
| Team Lead: | Sama Nikanfar |
| Other Team Members: | Aref Hebri | Harish Ram Nambiappan | Advitya Garg |
| Faculty Advisor(s): | Farnaz Farahanipad and Fillia Makedon |
| Department: | CSE |
| Project Type: | Graduate Student |
| Title: | Development of a Vision-Based Smart Cane with RGB-Only Perception for Collision Avoidance and Scene Recognition |
| Abstract: | This project proposes a novel vision-based smart cane that performs both collision avoidance and scene recognition. The smart cane is designed to be lightweight and user-friendly, relying solely on a webcam for obstacle detection and scene recognition. The RGB images from the webcam are used to perform obstacle detection and distance estimation for the collision avoidance task, and vision language models are used on the RGB images to perform scene recognition. Once an obstacle is detected at close distance or when the user needs to know the details about the environment, the smart cane provides feedback using vibration and audio, thus allowing users to avoid collisions and know the environment during navigation. Experiments were conducted to mainly perform a technical system evaluation of the smart cane, in terms of latency and accuracy, in indoor environments. The experimental results showed that the smart cane performs efficiently with only a single webcam in a lightweight design, along with basic haptic and audio feedback components in an indoor environment. |
| Poster: | View Poster #14 (PDF) |
| Judges Scoring: | Submit Evaluation (judges only) |