ANVAYA: AI-powered Navigation and Vision-based Assistive System for the Visually Impaired

Abstract

Abstract

Visual impairment significantly affects an individual’s ability to navigate safely due to limited perception of surrounding obstacles and environmental conditions. Traditional mobility aids such as white canes provide reliable ground-level obstacle detection but lack forward-looking spatial awareness, while many existing assistive systems focus primarily on object detection without delivering effective navigation guidance. This paper presents ANVAYA, a real-time assistive navigation system that integrates computer vision and ultrasonic sensing through sensor fusion to enhance environmental perception. A lightweight deep learning model is employed for object detection, while ultrasonic sensors provide accurate distance estimation for nearby obstacles. The proposed system introduces a zone-based spatial risk assessment model combined with a priority-driven decision framework to generate real-time directional guidance. Based on the assessed risk level, an adaptive multimodal feedback mechanism delivers context-aware audio and haptic alerts, enabling users to respond effectively while reducing cognitive load. The system is implemented on an embedded platform to ensure portability, low-latency operation, and practical usability in real-world environments. Experimental evaluation demonstrates improvements in obstacle detection accuracy, navigational responsiveness, and user safety, highlighting the potential of ANVAYA as an effective assistive solution for individuals with visual impairments.