Authors - Komal Papanwar, Selina Shrivastava, Anjali Askhedkar Abstract - This paper presents an intelligent assistive navigation system designed to enhance the mobility and spatial awareness of visually impaired individuals. The system is built on a compact and low-power Raspberry Pi platform, enabling portability and affordability. It integrates real-time object detection using the YOLOv8 deep learning model with AprilTag-based localization to estimate the user’s position within a mapped environment. A calibrated monocular camera captures visual input to detect both objects and tags, while geometrical calculations estimate distances and directional angles to guide navigation. Users can input their desired destination via speech, and the system provides auditory instructions using a text-to-speech engine. By combining computer vision, lightweight edge computation, spatial localization, and audio feedback, the system transforms visual data into actionable navigation commands. Experimental evaluations in controlled indoor environments demonstrate the system’s ability to deliver accurate, real-time guidance. This work advances the development of portable, intelligent, and cost-effective assistive technologies for independent mobility among the visually impaired.
