本研究之長程目標為發展具有高操控性的無障礙行動輔具，目前世界各國已有許多國家開始研究能夠方便使用者控制的助行器與機器手臂，並組合成可以行走與活動的機器骨骼系統。此系統未來在軍事上可以減輕士兵負重，進而可以提升士兵的作戰能力；在醫療層面上可以幫助肢體障礙的病人利用微弱的肌電訊號，進行精確的肢體行動、自主行動、自主照顧等，改善肢體障礙病人因肌肉無力所導致的日常生活不便。 藉由肌肉收縮時，擷取肘關節的肌電訊號，當作估測肌力的主要參數。由於肘關節兩側肌肉收縮產生的肌力與所量測的肌電訊號為非線性關係，且肌肉長度與收縮速度亦會影響所產生肌力。故本研究用肌電訊號與肘關節角度以及角速度當作倒傳遞類神經的輸入參數，代入倒傳遞類神經的訓練過程求得類神經內部權值，以準確的估測肘關節肌力。 The long term goal of this research is to develop the highly manipulated and accessible device. Until now, there are many countries in the world have started to research the exoskeleton system which will facilitate the daily activities of the disables. This device can be used in the military in the future to reduce the burden of soldiers and improve the operational capability. From medical perspective, the system can also assist physical disabled patients by accessing the feeble electromyographic signal data to support their physical operations, autonomous actions and improve their quality of daily activities. The electromyographic signal data measured from the contraction of the joint and muscle is used as the main parameter for estimate the joint torque. Considering the relation between the muscle strength result from the contractions of the triceps and biceps and the measured electromyogrphic signal is nonlinear, plus the muscle fiber length and the muscle contracted velocity also affect the elbow torque. Therefore, this research will use the electromyographic signal data, joint degree, and joint angular velocity as the input parameter, substitute the training steps to evaluated the weighting value of the backpropagation neural network to precisely estimate the joint torque.
