GigaHertz: Gesture Sensing Using Microwave Radar and IR Sensor with Machine Learning Algorithms

Intelligent systems have been important in all the industries whether its automotive, medical, automation and smart switches (in our terms replacing traditional switches). The key element of such systems is the user interface that can keep track of human gestures. Hand gesture sensing using a traditional approach i.e. camera has a serious limitation of lightning condition which triggers the use of additional infrared camera making the entire system bulky costly and less secure toward hacking problems. Short-range radar in combination with an IR sensor can sense all those gestures that a camera can by overcoming the drawbacks of the existing traditional approach. We have proposed a novel method to overcome the drawbacks of the existing systems and to use the proposed system GigaHertz for multiple applications, this method is based on the use of Doppler shift using radar and IR sensor for end to end (hardware, interface, and software) radar-IR-based system design to efficiently detect the gesture. The novelty of the proposed method is the collaboration of a radar system with an IR sensor to provide a total of eight gestures which can be configured as per user needs. The main advantage of the proposed system is in the selection of effective and efficient feature selection that enhances the classifier accuracy. A unique algorithm that we have designed to sync the various components and to configure different gestures as per user needs and also achieve high accuracy of 92% with SVM machine learning algorithm which is higher than existing gesture recognition technology. We aim at replacing the traditional switch with the proposed model device making it safer to use for multiple applications controlling a wheelchair, home automation, to control various systems in a vehicle making it safer for the driver to drive without taking eyes off the road, controlling quadcopter and many other such applications.