Telepresence Avatar Motion

How can we understand users’ intentions in one place and intelligently transmit their corresponding movements with the same meanings to avatars placed in different spaces? Telepresence Avatar Motion research aims to enable avatars to reproduce users’ behavioral characteristics as they interact with spaces, objects, and other people in real time. This novel approach can be applied in virtual meetings, remote collaboration, and immersive communication systems where natural expression is essential.

Our current research focuses on understanding and transmitting users’ upper-body motion intentions through neural network-based learning approaches. We are developing technologies to accurately analyze and reproduce gestures expressed through head and hand movements. In the future, we plan to extend this technology to full-body motion, enabling complete physical expression, including walking patterns, posture changes, and leg movements. By progressively expanding our deep learning methods from upper-body to full-body motion, we aim to build an intelligent virtual communication system that enables avatars to convey users’ intentions more naturally and intuitively.