Fusion of Object Recognition and Obstacle Detection approach for Assisting Visually Challenged Person

Abstract

An assistive device is proposed for visually impaired people to provide them automatic navigation and guidance. The device can impart navigation free of obstacles and perform real-time image processing to acknowledge the obstacle. The method is intended to perform all tasks in real-time to get all the information about obstacles in the path with optimised information flow. The proposed navigation stick is low cost obstacle detector, obstacle identifier and object detector. The whole setup for rehabilitation for visually impaired people consists of the heterogeneous set of sensors and computation components such as - ultrasonic sensors, camera, single-board DSP processor, a wet floor sensor, and a battery. The machine learning model is used for object recognition to make the user familiar with the environments. To provide the best navigation guidance, proposed device can be used to detect obstacles, up-stairs, down-stairs, edges, potholes, speed breaker, narrowing passages, pothole, wet floors, etc. The output is provided in the form of the audio prompt instead of the vibration, to make user comfortable and user-friendly.