AI-Powered Visual Navigation System

Abstract

The Augmented Reality (AR)Navigation System for Visually Impaired Individuals aims to enhance their ability to move independently in unfamiliar environments. Using a mobile device\'s camera, the system captures real-time images, processes them to identify objects. Users interact with the system through voice commands, enabling hands-free operation, while synthesized speech delivers navigation instructions and object descriptions. This approach helps visually impaired individuals by offering real-time, on-demand access to visual information, improving their mobility and independence. The system leverages Al, image processing, and voice interaction for an efficient and accessible navigation solution.

Introduction

Overview

Position tracking is essential for modern applications, but GPS fails indoors—a major issue for visually impaired individuals. To address this, an AI-powered visual navigation system has been developed, which uses a device’s camera, AI (YOLOv5), and text-to-speech technology to provide real-time auditory guidance for safe and independent mobility.

Key Features

• Object Detection: Uses YOLOv5 to identify people, vehicles, furniture, and obstacles.

• Speech Output: Converts object data into voice feedback using pyttsx3, providing directional cues like “Move right.”

• Voice Commands: Incorporates speech recognition to allow hands-free operation via voice input.

• Indoor & Outdoor Use: Works across varied environments like buildings, streets, or parks.

Related Work (Literature Survey)

• Mixed Reality Gloves (Omary & Mehta, 2024): Enable visually impaired users to interact with virtual objects through touch and audio cues.

• Indoor Navigation (Croce & Giarre, 2019): Uses phone cameras and floor pattern detection for indoor movement with vibration feedback.

Existing Systems

• Microsoft HoloLens 2: Uses spatial mapping and 3D audio feedback (A3DF) to help users navigate by hearing directional cues.

Proposed System

• Combines YOLOv5 object detection with speech synthesis and voice commands to create a robust, real-time guidance system.

• Converts camera input into meaningful voice alerts, improving situational awareness and navigation safety for visually impaired users.

Modules

1. Object Detection – YOLOv5 with 0.4 confidence threshold.

2. Speech Recognition – Converts voice commands to actions using Google’s API.

3. Text-to-Speech – pyttsx3 speaks out object locations and navigation tips.

Benefits

• Hands-Free & Real-Time: Operates continuously without user input.

• Adaptive: Responds to dynamic environments.

• Safe & Informative: Guides users with spoken instructions.

• Cost-Effective: Runs on standard mobile devices with cameras.

Conclusion

The AI-Powered Visual Navigation System is an innovative and supportive technology developed to aid visually impaired individuals in navigating their surroundings safely and independently. At the heart of this system is a camera that continuously captures real-time images of the environment. These visuals are processed using advanced artificial intelligence algorithms capable of recognizing and analyzing objects, obstacles, pathways, and other critical spatial details. The AI interprets this information instantly and translates it into clear, spoken instructions that guide the user step by step.

References

[1] WafaM.Elmannai,KhaledM.Elleithy. “A Highly Accurate and Reliable Data Fusion Framework for Guiding the Visually Impaired”. IEEE Access 6 (2018) :33029-33054. [2] Qi-Chao Mao,Hong-Mei Sun,Yan-Bo Liu,RuiShengJia. “MiniYOLOv3:Real-Time Object Detector for Embedded Applications”.IEEE Access 7 (2019) :133529 133538. [3] ZhenchaoOuyang,JianweiNiu,YuLiu,MohsenGuiz an i. “Deep CNN-Based Real-Time Traffic Light Detector for Self-Driving Vehicles”.IEEE Access 19 (2019):300-313. [4] Wei Fang,LinWang,Peiming Ren. “Tinier YOLO:A Real-Time Object Detection Method for Constrained Environments”.IEEE Access 8 (2019) :1935-1944. [5] Meimei Gong,YimingShu. “Real-Time Detection and Motion Recognition of HumanMovingObjects Based on Deep Learning and Multi-Scale Feature Fusion in Video”.IEEE Access 8 (2020):25811- 25822. [6] VidulaV.Meshram,KailasPatil,VishalA.Meshram, Fe lix CheShu. “An Astute Assisstive Device for Mobility and object Recognition for Visually Impaired People”.IEEE Access 49 (2019) :449 460.

Copyright

Copyright © 2025 Mrs. Md. Jareena Begum , S. Subba Lakshmi , A. Prashanthi, Ch. Lakshmi Rohitha, G. B. S. P. Vaishnavi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.