A Real-time Full-body Motion Capture and Reconstruction System for VR Basic Set

PubDate: April 2021

Teams: Shanghai University

Writers: Linhui Gu; Liwen Yin; Jinyao Li; Di Wu

PDF: A Real-time Full-body Motion Capture and Reconstruction System for VR Basic Set

Abstract

Having a virtual avatar in virtual reality(VR) applications can increase immersion. However, most of the existing VR basic sets only contain limited tracking sensors to provide effective information for full-body motion capture and reconstruction. Therefore, we present a lightweight real-time full-body motion capture and reconstruction prototype system only using head-mounted display and motion controllers. The system constructs HeadIK and HandIK which are optimized based on the human inverse kinematics (IK) algorithm to reconstruct the upper body motion accurately. And we address the animation blending space and FootIK to achieve natural smooth lower body motion that is not neglected by blindfolded users. Meanwhile, the animation retargeting algorithm ensures that our system can adapt to various body shapes. Experiments show that our method can reconstruct the natural movement of the whole body using only consumer-grade VR basic sets. We demonstrate that our system can enhance immersion in the VR experience and can be used in most VR applications.