整合機器學習於線纜式頸部角度量測裝置

Abstract

頭頸部的活動度會影響與他人及環境互動的能力，也因此導致控制頸部運動上有困難的族群，如低頭綜合症患者，他們的獨立生活能力降低。傳統輔具多採用固定式裝置確保頸部的安全，但此靜態拘束仍舊限制了互動能力並常因為長時間穿戴下肌肉緊繃產生不舒適感，因此有必要開發可協助頸部肌肉控制動態活動且穿戴舒適的頸部輔具。為確保所採用機構不影響頸部的活動範圍，本研究利用並聯式線驅動機構設計軟性頸部穿戴裝置，使用渦卷彈簧作為集線器並提供線張力以輔助肌肉維持不同頸部姿勢與運動，為量化輔助程度，在頸部運動實驗中以肌電訊號評估肌肉激發強度的變化，結果顯示在抬頭時，頭夾肌在穿戴裝置下的激發強度比未穿戴小19.7%，可得裝置往後的線拉力確實減低支撐頭重的負擔。為量測頸部旋轉角度以提供醫師診斷與治療的參考，集線器加入角度感測的磁編碼器以藉由集線器旋轉圈數來計算機構中每條線長的變化，在使用頸部運動的量測結果進行機器學習後獲得運動學模型，使用未加入模型訓練的數據進行驗證得到方均根誤差值為1.68°。總體而言，本研究所開發的裝置可提供被動輔助與量測頭頸部旋轉角度，未來可朝提供主動式線張力控制發展以提供更有效的頸部運動輔助。

The movement of the head-neck plays a crucial role in interacting with others and the environment. Thus, individuals with neck movement control difficulties, such as patients with dropped head syndrome, often have lower independent living ability. Traditional assistive devices are mostly static collars to ensure neck safety. However, such static constraints limit interaction capabilities and often lead to discomfort due to prolonged muscle tension. Hence, there is a need to develop neck braces that assist in dynamic movement control while being comfortable to wear. To ensure that the chosen mechanism does not hinder the range of neck motion, this study designs a soft neck brace using a cable-driven parallel mechanism. A spiral spring is employed as a spool to provide cable tension, assisting muscle activation. Electromyography was used to assess muscle activation levels during neck movement experiments. The results indicated that during head extension, the activation level of the splenius capitis muscle was reduced by 19.7% when wearing the device, demonstrating that the posterior cable tension of the device effectively supports part of the head's weight. For measuring neck rotation angles to aid in diagnosis and treatment, the hub was integrated with a magnetic encoder for spool rotation angle measurement to obtain cable length. A machine learning-based model was derived based on cable lengths and neck movement measurements. Validation yielded a root mean square error (RMSE) of 1.68°. Overall, the device proposed in this study can provide passive assistance and measure head-neck rotation angles. Future works could focus on active neck movement assistance.