建構一套基於深度攝影機和重心回饋之太極拳訓練系統

Abstract

相較於傳統的多媒體影片的學習方式，本研究建構一套基於深度攝影機及重心回饋科技來提供視覺引導和學習回饋之智慧型太極拳訓練系統。進而讓使用者可藉由本系統在沒有教練在的情況下獨自練習。為使得動作訓練系統更加完善，本研究針對深度攝影機的諸多限制，整理並提出相關系統之設計準則，以讓系統可獲取更正確的資訊來做動作分析。另外，在多媒體學習策略部分，本系統採用七階層漸進式學習模組，其中結合了動作學習和重量轉移學習兩部分。最後，本研究採用三種不同介面來評估本系統－分別是只看影片學習、顯示骨架的引導和回饋學習，與顯示骨架和重量轉移的引導和回饋學習。結果顯示有加入重量轉移的引導和回饋機制對於學習太極拳動作非常有效，且重量轉移的學習可幫助學員在肢體動作上的學習。

In comparison with the traditional learning method by video, our research is developing a smart physical training system to practice Tai-Chi Chuan with a depth-camera-based and pressure sensing technology for supporting visual guidance and feedback. Users are capable of practicing TCC alone without instructors. To make our training system be more wonderful, we focus on investigating the limitation caused by a depth-camera-based, and advance the design guidelines for acquiring more correct information to analyze. Moreover, in our learning module, we introduce seven step-by-step learning stages which include movements learning and weight-transfer learning. Finally, our research adopts three different interfaces to evaluate our training system, which are video only, skeleton feedback, and skeleton&weight-transfer feedback interface. The result demonstrates that adding weight-transfer feedback is beneficial for trainees to learn the Tai-Chi movements.