Towards Controlling Whole Body Avatars with Partial Body-Tracking and Environmental Information

PubDate: April 2022

Teams: Utsunomiya University

Writers: Koji Yamada; Hiroshi Mori; Fubito Toyama

PDF: Towards Controlling Whole Body Avatars with Partial Body-Tracking and Environmental Information

Abstract

In body-tracking-based avatar manipulation, the user’s motion is reflected in their avatar, which creates a high level of immersion. Conversely, the user is required to give a detailed performance similar to that of the avatar to be composed. In this research, we aim to produce avatar motion that reflects the user’s intended actions and maintains consistency with the VR environment by inputting the user’s posture. The avatar’s body motion was generated by inputting the user’s motion and VR environmental information into the motion configuration network, which was developed using machine learning.