fairfolio
Saved Projects

A Multimodal Edge-AI Smart Cane for Real-Time Fall Prevention

CSEF · 2026 Behavioral & Social Sciences (Senior Division)

Alex Zhang

Overview

Falls are a leading cause of injury-related hospitalization and loss of independence in adults aged 65 and older, yet most assistive systems remain reactive, detecting falls only after impact. This project developed a proactive, multimodal edge-AI smart cane that identifies environmental trip and slip hazards in real time before foot placement. The engineering goal was to design a portable, low-latency system capable of detecting common walking hazards during natural cane use. The cane integrates a Raspberry Pi 5, Raspberry Pi AI HAT+ 2, wide-angle camera, 9-DoF IMU, and speaker into an embedded platform. A sliding-window gait-stability algorithm triggered image capture during low-motion walking intervals to reduce motion blur. Images were processed fully on-device using edge-compatible vision-language models, while temporal filtering and sensor fusion were used to reduce false alerts. Audio warnings were issued when hazards such as cords, rug edges, or wet surfaces were detected. Using more than 500 cane-acquired images, five vision-language models were benchmarked for precision, recall, F1 score, and inference latency. MoonDream2-2B achieved the best overall performance with precision 0.86, recall 0.94, and F1 score 0.90. Resolution testing showed that QVGA reduced average latency to under 2.0 s, about 2× faster than higher-resolution conditions, with only modest loss in detection accuracy. These results indicate that low-cost, fully offline edge-AI hazard detection is feasible on a smart cane and can provide practical real-time warning before a fall occurs.

Competition history

  • CSEF 2026 Behavioral & Social Sciences (Senior Division) · Entry S-03-18

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: California Science & Engineering Fair public projects

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google