Visually Impaired People Navigation System using Sensors and Neural Network

Without a personal assistant or a trained dog, it is difficult and somehow risky for visually impaired individuals to navigate indoors and outdoors. Navigation for individuals with visual impairment is now an active area of research, and numerous systems have been created. Developing a reliable and robust system will give the visually impaired more awareness about their surroundings, provides more independence in their lives, and reduces the risks they might face. This paper proposes the design and implementation of a prototype for a wearable system that enables blind and visually impaired individuals to navigate easily and safely by using computer vision and deep learning to identify objects and sensors to detect obstacles. Using an attached camera, the system uses image processing to recognize objects around the user. It identifies the objects based on a trained deep learning dataset and audibly communicates the object names to the user. Ultrasonic sensors detect nearby obstructions and alert the user by vibrating. Upon request, a Global Positioning System (GPS) receiver transmits the user’s location via Short Message Service (SMS) to a caregiver. All the sensors and devices are connected to and controlled by an Arduino Mega microcontroller that has been programmed with efficient and dependable algorithms to perform tasks precisely and rapidly. Object recognition and reading are performed using a computer equipped with the necessary MATLAB software and libraries.