Consumer-Level Tai-Chi Teaching Assistance System: Feedback-Based Teaching of Traditional Chinese Sports in University Environments Integrating Visual Information and Virtual Reality
Wensu Zhang, Shiyan Zhang, Kun Ma · International Journal of Human-Computer Interaction · 2025
To overcome limited feedback and personalization in large university Tai Chi classes, we build a VR-based teaching system that pairs real-time visual interaction with Unity3D. Depth cameras track students’ skeletons, while Dynamic Time Warping and Hausdorff distance compare their trajectories to expert templates. A Kalman filter smooths data, and a decision-tree classifier instantly grades each move, delivering individualized cues and progress reports. The closed-loop design lets learners repeatedly refine posture, timing, and flow, boosting motivation and accuracy. Because algorithms and content are modular, the platform scales to other sports and large cohorts without added instructors. Pilot trials with 66 undergraduates demonstrated significant gains in form accuracy and self-reported engagement versus traditional instruction. The framework supports asynchronous practice, remote coaching, and multimodal analytics that can inform curriculum design. Future work will integrate haptic feedback and adaptive difficulty to further personalize instruction.