SmartAssist Vision is a wearable assistive technology designed to enhance independent mobility for visually impaired individuals through real-time object detection, obstacle awareness, and voice-guided navigation.
The system combines Computer Vision, Edge AI, Distance Sensing, and Speech Feedback to provide situational awareness without requiring internet connectivity.
Traditional mobility aids provide limited information about surrounding objects and obstacles.
SmartAssist Vision addresses this challenge by:
- Detecting objects in real time using MobileNet-SSD
- Identifying object position (Left, Center, Right)
- Measuring obstacle distance using an HC-SR04 ultrasonic sensor
- Providing voice instructions through earphones
- Running entirely on a Raspberry Pi 4
- Real-Time Object Detection
- Voice-Guided Navigation
- Obstacle Alert System
- Distance Measurement
- Offline Edge AI Processing
- Portable Wearable Design
- Low-Cost Implementation
- Raspberry Pi 4
- USB Camera
- HC-SR04 Ultrasonic Sensor
- Earphones / Headphones
- Power Bank
- Wearable Helmet Mount
- Python
- OpenCV
- MobileNet-SSD
- NumPy
- RPi.GPIO
- eSpeak
- Raspberry Pi OS
- Camera captures surroundings
- MobileNet-SSD detects nearby objects
- Position logic identifies object location
- Ultrasonic sensor measures obstacle distance
- Raspberry Pi processes sensor data
- Voice feedback is delivered through earphones
- Camera Validation
- MobileNet-SSD Testing
- Object Position Detection
- Voice Output Testing
- Raspberry Pi Integration
- Ultrasonic Sensor Integration
- Distance Classification
- Voice Feedback System
- Complete Wearable Deployment
- Person on the left
- Chair in center
- Bottle on the right
- Caution table very close
Camera → MobileNet-SSD → Object Detection
HC-SR04 → Distance Measurement
Raspberry Pi → Data Fusion
eSpeak → Voice Guidance
- Assistive Technology
- Smart Mobility Systems
- Wearable AI Devices
- Edge AI Solutions
- Accessibility Research
- GPS Navigation Assistance
- Emergency SOS Feature
- Face Recognition
- AI Scene Understanding
- Mobile Application Integration
- Coral TPU Acceleration
Fardin Imran Shaikh
This project is intended for educational and research purposes.