AIoT Blind Stick Based Independent Navigation with Enhanced YOLO Object Detection
S V Hariprasad, N. Bharathiraja, Thangavel Deepa, R. K. Nehaa, Daniel Joseph Samuel, S Navya Sri · 2024
Nowadays, Artificial Intelligence (AI) and the Internet of Things (IoT) herald a transformative leap in assistive technologies, epitomized by the AIoT Blind Stick. This smart mobility aid the computational prowess of the Raspberry Pi to process a multitude of sensory inputs, delivering realtime environmental awareness to visually impaired users. The system provides auditory and haptic feedback to navigate complex urban landscapes safely and is equipped with GPS for location tracking, ultrasonic sensors for obstacle detection, and a camera interfaced with advanced object detection algorithm using YOLO8. Voice Based assistance given based on the object detected at each frame. The proposed YOLOv8m model demonstrates superior detection performance, achieving the highest Average Precision (AP) of 50.4 among the evaluated models. Although the frame rate (FPS) of 137 is slightly lower than that of the smaller models (YOLOv5s and YOLOv8s), it provides a significant balance between accuracy and processing speed, making it highly suitable for real-time object detection applications where precision is critical, such as assistive technologies for the visually impaired.