Assistive Stick for Visually Impaired People

Blind or visually impaired people have difficulty navigating the street and are also unaware of nearby landmarks. They need an assistant (e.g. a cane, human, dog, e.t.c.) to go from one place to another place. In this modern era, it is possible to develop smart solutions with the help of modern technologies like deep learning (DL), natural language processing (NLP), internet of things (IoT), etc. that can assist them without any human intervention. A smart stick is one of the solutions that can help them with navigation and get to know the surrounding places. This paper proposed a smart solution and developed a smart stick to help visually impaired people with the use of DL and NLP and devices like the Raspberry Pi, an ultrasonic sensor, and a camera. This smart stick can detect obstacles using an ultrasonic sensor within the range of 1.5 metres and alert the person by generating audio as a warning message. The stick is also able to capture images from its surroundings using the camera and can read the text present in the captured image. The text in the image is detected using a deep learning model called the EAST text detector and recognised using an OCR Engine tesseract. Furthermore, the recognised text is converted into audio so that the person can hear about their surroundings. All calculations and processing are done on the Raspberry Pi, and the generated audio can be heard by using a speaker(earphone or headphone). The proposed solution is successfully implemented and demonstrated, and it is found to be very efficient and cost-effective.