Independent mobility remains a major challenge for visually impaired individuals, particularly in dynamic and unfamiliar environments where obstacles and navigation hazards are unpredictable. Conventional assistive aids such as white canes and guide dogs offer limited environmental perception and do not provide real-time situational awareness. This work proposes a Real-Time AI-Based Navigation and Obstacle Detection System for Blind Users aimed at improving safe and autonomous movement in indoor and outdoor settings. The system combines computer vision, deep learning, and sensor fusion to continuously analyze the surrounding environment. A lightweight convolutional neural network processes live video input to identify both static and moving obstacles, while depth estimation and ultrasonic sensors are used to accurately measure distance and assess collision risk. An intelligent path planning module determines safe navigation directions based on real-time spatial analysis. Guidance is conveyed to users through audio cues and haptic feedback via wearable devices, ensuring intuitive and unobtrusive interaction. Edge-based processing enables low-latency responses and reliable performance even in low-connectivity conditions. Experimental results indicate enhanced obstacle detection accuracy, faster response times, and increased user confidence when compared with traditional mobility aids. The proposed system provides an efficient, affordable, and scalable solution to support independent navigation and improve quality of life for visually impaired users.
S et al. (Thu,) studied this question.