Embedded Design of Smart Helmet for Physically Impaired

The work proposed in this paper is a prototype for visually impaired people to identify common real objects, ensure their safety in public transportation and during panic situations. Here, we are to replace blind aided stick with Smart Helmet and integrating some features. This is carried over Embedded technology with the help of Machine Learning techniques. Computer Vision algorithm is implemented on Raspberry Pi detects Objects and warns the visually impaired person through speech in order to navigate properly. TensorFlow framework is used for real-time object detection in ML. This system has Panic Alert switch which the visually impaired can press when they are in an uncomfortable scenario. When a Panic Alert is triggered, our system sends the location, which was recorded using the GPS module, to the emergency contacts through GSM. GPS receiver will communicate with satellites and them it will calculate the latitude and longitude to track the location. Flite (Festival lite) is used to convert from text to speech. The primary goal of this work is to create a Helmet with convenient size and easy to navigation aid for Visually impaired which helps in Artificial Visualization by providing facts about the environmental scenario and forceful objects around them.