Research on Virtual Reality Arm Motion Capture and Recognition

Shubin Cai, Dihui Deng, Jianqiang Li, Jinchun Wen, Chaoqin Chen, Zhong Ming

Abstract: In current virtual reality (VR) applications, when you wave your hand with the VR controller, you will see a controller floating in the air and shaking from side to side in a VR environment. Or in some better situation, there is also a wave of your virtual arm. But the virtual arm is moving in a different way with your real arm or a higher delay, and you will feel that the virtual arm is clumsy. This situation seriously reduces the user immersion feeling that VR industry attach great importance to. There are systems that can quickly and accurately capture and replay arm motions nowadays, but they undoubtedly cost more than a normal VR device. Therefore, we need a set of fast, accurate and cheap arm motion capture, recognition and replay solution that can be used in VR environment. Based on the HTC VIVE platform, this paper proposed a much less expensive kinematics-based VR arm motion solution. The traditional dynamic time warping (DTW) algorithm is improved by feature constraint and feature matching order strategy. Experiments show that the new algorithm can improve the response speed by 36% and reduce the response time to 216 milliseconds while maintaining the accuracy of 95.6%. The whole arm motion solution can greatly improve users’ immersion experience in a VR environment.