動作調校服飾系統技術

Technology of Smart Textile for Human Posture Recognition

莊秉欽(P. C. Chuang) ； 賴允昱(Y. Y. Lai)

《紡織綜合研究期刊》 31卷3期 (2021/07) Pp. 62-69

https://doi.org/10.6439/TTRJ.202107_31(3).0008

摘要

本研究目的為開發動作調校服飾系統技術，協助運動者在運動過程中，得到即時的姿態辨識與監測，進而瞭解本身動作型態是否正確，以避免運動傷害發生，達到運動姿態調校之目的。方法：將九軸感測器與彈性導電迴路整合在衣服上，藉由運動行為產生之關節彎曲角度分析運動行為，建立動作研究演算法相關模式，分析系統的準確性，並據此修改演算法，加強產品之信度，以此建立較為便利的姿態辨識數據分析系統。實驗：招募一名具熟悉深蹲動作練習之受試者，穿著動作調校服飾，截取衣服內的感測器所得到的資料，錄製其深蹲動作曲線，並與標準正確曲線進行峰值對齊，再將兩個曲線丟入深蹲辨識演算法中進行運算，如果運算出來的誤差大於演算法定義的最大誤差則在客戶端顯示錯誤；小於等於則在客戶端顯示正確。根據研究結果顯示，當動作曲線超過10°的臨界值時，才會開始截取曲線，當動作曲線低於10°時，較無法經由演算法截取曲線。最後依受試者錄製20筆正確與15筆錯誤的資料，以此跟標準正確曲線進行比較結果，得到的演算法正確方準確率為80％，錯誤方準確率為100％。其中正確方準確率僅有80％的原因，主要為九軸感測器在感測角度上所存在的誤差所導致，建議動作調校服飾未來在開發與設計上，需要針對此問題加以克服。

關鍵字

九軸感測器 ； 深蹲動作 ； 姿態辨識

並列摘要

The technology is based on the needs of the next generation of performance sportswear consumers. In the process of exercise, it is essential to acquire correct posture recognition and comfort wearing. The smart textile is applied to the analysis and monitoring of sports postures, to help athletes understand whether the movement pattern is correct, and to develop preventive systems and achieve the purpose of sports posture adjustment. According to the results, the data will be captured when the action angle is greater than 10°. When the action angle is less than 10°, the data will not be captured. Finally, there are 20 correct and 15 incorrect data from the subject action records, and then compared with the standard correct curve. Through algorithm, the detection accuracy rate of the correct action was 80%, and the detection accuracy rate of the error action was 100%. The correct motion detection accuracy rate was only 80%, mainly due to not enough of the nine-axis sensors in the sensing angle. In the future, it is recommended that the sensing angle insufficient should be overcome of smart textile human posture recognition development and design.

並列關鍵字

9-axis sensors ； squat ； gesture recognition