Staircase Detection to Guide Visually Impaired People: A Hybrid Approach

Eyes and visionary organs are an essential part of human physiology, since they are capable of receiving and processing subtle details to the brain. Some individuals are not capable to visually perceive things around the environment. They face various types of hindrances such as obstacles, potholes, staircases, pedestrians and vehicles in the daily life and are not able to navigate themselves without a guidance. This study aims to help visually impaired people to navigate around the surroundings. A hybrid approach was developed for the detection of staircase and the ground using a pre-trained model and an ultrasonic sensor. In the proposed system, staircase images are captured via an RGBD camera and compared with pre-trained template images for detection. The developed system that employs an ultrasonic sensor, an RGBD camera, a raspberry pi and a buzzer is installed on a stick. Under a variety of conditions, the proposed system was tested using different stair images and achieved an accuracy of 98.73% in average. The system works well under various conditions, such as dark and noise.