Scholarlink

<p>Crossing a street, avoiding an electric pole, or detecting a door are routine tasks for most people. However, for individuals with visual impairments, such actions can pose serious challenges that compromise their safety and mobility. Assistive technologies have increasingly played a vital role in improving the quality of life for these individuals by promoting greater independence in daily activities. In this paper, we present the Road Obstacle Detection Device (RODD), an AI-based, edge computing wearable system designed to assist visually impaired users in navigating outdoor environments by detecting and identifying road obstacles in real time. RODD is a wearable, hands-free device that includes a Raspberry Pi, a smart camera, and ultrasonic sensors. This setup enables effective detection of both low-lying and elevated obstacles without requiring internet connectivity or interfering with the user’s natural movement. The device utilizes the YOLO (You Only Look Once) object detection model, trained on our custom <i>Road Obstacle</i> dataset, which includes 14 obstacle classes and over 12,000 annotated instances. The system demonstrates strong performance, achieving a precision of 0.82, a recall of 0.821, and an mAP@50 of 0.861. These results suggest that RODD is a promising tool for enhancing safe and independent mobility for the visually impaired.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assistive Device for Visually Impaired Individuals Featuring Road Object Detection

  • 期刊论文

  • 发表日期：

    2026年1月20日

查看全文
  • Imane Nedjar,
  • Mokhtaria Bekkaoui,
  • Lotfi Fazil Boukli Hacene,
  • Mohammed M’hamedi,
  • Fatma Bedaif,
  • Mohamed Amin Benzineb

摘要

Crossing a street, avoiding an electric pole, or detecting a door are routine tasks for most people. However, for individuals with visual impairments, such actions can pose serious challenges that compromise their safety and mobility. Assistive technologies have increasingly played a vital role in improving the quality of life for these individuals by promoting greater independence in daily activities. In this paper, we present the Road Obstacle Detection Device (RODD), an AI-based, edge computing wearable system designed to assist visually impaired users in navigating outdoor environments by detecting and identifying road obstacles in real time. RODD is a wearable, hands-free device that includes a Raspberry Pi, a smart camera, and ultrasonic sensors. This setup enables effective detection of both low-lying and elevated obstacles without requiring internet connectivity or interfering with the user’s natural movement. The device utilizes the YOLO (You Only Look Once) object detection model, trained on our custom Road Obstacle dataset, which includes 14 obstacle classes and over 12,000 annotated instances. The system demonstrates strong performance, achieving a precision of 0.82, a recall of 0.821, and an mAP@50 of 0.861. These results suggest that RODD is a promising tool for enhancing safe and independent mobility for the visually impaired.