Blind Lane Detection and Following for Assistive Navigation of Vision Impaired People

Abstract

The task of safely navigating outdoor environ-ments presents significant difficulties for people with visual impairments. This paper aims to address this issue by proposing a wearable assistance system for visually impaired individuals in blind lane detection and following scenarios. The system consists of two stages: the first stage employs a fully convolutional network to detect blind lanes and an object detection network to identify obstacles. The second stage utilizes an improved artificial potential field method to achieve real-time path following based on the detection results. Our experiments involving subjects in dynamic outdoor environments demonstrate the robustness of our proposed method in navigating and avoiding obstacles under challenging outdoor conditions.