Design of Blind-Assist Smart Navigation Goggles Using Raspberry Pi

Abstract

This paper presents the design and implementation of Blind-Assist Smart Navigation Goggles, a wearable assistive device empowering visually impaired individuals to navigate independently. The goggles integrate Raspberry Pi 4 as the central processor, a 5MP camera for real-time environment capture, HCSR04 ultrasonic sensor for obstacle proximity detection, and YOLOv8 deep learning model for accurate object and obstacle detection. Detected objects (persons, vehicles, furniture, steps) are communicated through natural language voice alerts via text-to-speech engine. The ultrasonic sensor runs in parallel at 15 Hz for close-range obstacle alerts. The system achieves end-to-end detection-to-voice latency of 1.2 seconds with 88.5% overall detection precision across 15 object categories, providing practical real-time navigation assistance for visually impaired users.