UWB/IMU-assisted Gesture Recognition Using Learning Approaches for VR/XR Applications

In this paper, to efficiently support virtual reality (VR) and extended reality (XR) applications, we propose a novel method for gesture recognition utilizing inertial measurement unit (IMU) and ultra-wideband (UWB) in the commercial off-the-shelf, where three smartphones are used for one head mounted display (HMD) and two wrist-attached devices (WADs). According to the intensive experimental evaluations, our proposed method can provide very high success rate of gesture recognition for pre-defined 10 hand motions. For 8 volunteers participating the data collection, classifier learning, and testing, we observed that the recognition success rate of 100 % is achieved. Moreover, the success rate of 91.9% for new users who did not participate in learning can be achieved. These results can provide meaningful guidelines to adopt the UWB chipset into the HMD and WADs, because UWB significantly improves the success rate of gesture recognition than that result from the IMU only.