Navigation System for Blind People Using Machine Learning

Abstract

Proper navigation and detailed perception in familiar or unfamiliar environments are the main roles for human life. Eyesight sense helps humans to abstain from all kinds of dangers and navigate to indoor and outdoor environments. These are challenging activities for blind people in all environments. Usually, they are navigated to their natural sense but sometimes this is not adequate because human senses can be affected by external noise. Many assistive tools have been developed by the blessing of technology like white canes, braille compasses that help blind people to move around the world. In this project, we have developed a cloud and vision-based navigation system for visually impaired people or the blind. Our aim is not only to navigate them but also to perceive the environment in as much detail as a normal person. The proposed system includes ultrasonic sensors detecting obstacles, stereo camera to capture videos to perceive the environment using deep learning algorithms. The proposed system also includes a face recognition system that identifies known faces in front of him. Blind people interact with the whole system through a speech recognition module and all the information is stored in the cloud. Web application and Android application are used to track blinds, visited areas, etc. For that, the guardian is able to monitor them all the time. Blind is also contact their guide in an emergency situation. The experimental results show the proposed system can provide more abundant surrounding information and interaction with the newly proposed system.