AI-Enabled Real-Time Environmental Awareness and Guidance System for Visually Impaired Users

Abstract

Safe and independent navigation is also one of the biggest challenges for visually impaired people, especially in dynamic environments both indoors and outdoors. Existing assistive aids have a low level of contextual awareness, not to mention real-time decision-making capabilities. This paper presents an AI enabled real-time environmental awareness and guidance system to provide an intelligent navigation assistance by computer vision and audio feedback. The proposed system uses a pretrained YOLOv8 object detection model to detect the obstacles and the environmental elements from live video streams. Direction and distance estimation techniques are employed to find the position of the obstacles while a decision based on rules is used to prioritize risk and formulate navigation instructions. The system offers clear, multilingual audio guidance to assist obstacle avoidance, vehicle warnings and identifying the path is clear. Experimental evaluation shows reliable real-time performance in terms of satisfactory detection accuracy, fast infer speed and high path clearance accuracy. The proposed approach provides a low-cost, scalable, and practical assistive solution that promotes environmental awareness, safety, and independent mobility of the visually impaired users.