A Robust Real-Time Multiuser Gesture Recognition System for Human–Computer Interaction Using mmWave Radar Sensors

Abstract

Human gestures are widely used as a control interface between the user and the device. Camera and infrared-based sensors offer a state-of-the-art solution for human gesture recognition. However, these dominant sensors have raised concerns related to privacy and functioning in different environments. Concurrently, radar sensing technologies are maturing and hold a notable advantage in addressing these issues. Despite this progress, situations may arise where multiple users need to interact simultaneously with the same system, posing challenges for current radar-based solutions. In this article, we introduce a multiuser gesture recognition system for human–computer interaction using a millimeter-wave (mmWave) radar sensor with our own novel points’ clustering method. Our proposed system can simultaneously track and recognize the gestures for two users, achieving an average accuracy of 92.80% for five gestures. A pilot study is also conducted to evaluate the feasibility of the system.