本研究提出應用智慧體感網路資訊推論技術，以實現一個雲端高爾夫運動學習系統。該技術包括擊球結果推論與動作姿勢修正，主要利用四元數（Quaternion），設計一球杆-地球全域座標間的感測資訊轉換方法，藉此計算出正確的揮杆速度、節奏、軌跡等資訊。再參考多裝置感測器(如球杆、腰間智慧型手機等)間的資訊，研發擊球結果推論方法，判斷是否擊出失誤球，如左曲球（Hook）、右曲球（Slice）、左拉球（Pull）、右推球（Push）、打地球（Duff）和剃頭球（Top）。最後同樣藉由多裝置感測器資訊，綜合判斷出揮杆結果不理想的主要原因。經過實際於高爾夫練習場揮杆測試與專業教練指導，所提出之擊球結果推論與動作姿勢修正建議的正確率均達到的83%以上。

In this paper, we propose intelligent body-sensor-network information inference techniques for cloud-based golf sport training system. The techniques mainly include swing result inference and posture correction. A coordinate transform of sensor-frame and earth-frame is designed for calculating correct swing results, e.g., speed, tempo and trajectory. Besides, according to multiple sensor information, a proposed method of swing result inference can determine whether hit missing balls or not, e.g., hook, slice, pull, push, duff and top balls, and give beginning players suggestions for posture corrections. In our experiments, with a professional coach's advising, the developed sport training system can achieve 83% accuracy for swing result inference and posture correction.