肌電圖強度與速度分析於機器手臂控制之應用

Abstract

EMG(electromyography)訊號是肌肉在活動中收縮而產生的一種類比生理訊號，因使用者的意圖而有不同的特徵，因此利用肌電圖作為義肢或機械手臂的控制命令，是相當直接及自然的方法。基於此，我們針對肌電圖與因應的手部動作以及速度辨識進行分析，建構出一套及時的基於肌電波機械手臂控制系統，操作者可以手肘的伸或曲，快或慢來操作機械手臂，此系統利用2-channel 的表面電極從肱二頭肌與肱三頭肌得到肌肉訊號，接著在時域及頻域上評估肌肉動態及強度，透過分類器辨識及速度曲線映射辨識出前臂之動作與大約的速度，再轉成機械手臂之控制命令，機械手臂則根據接收之指令做出相對應的動作，為了達到即時控制之目的，我們也發展一套訊號整合分析及機械手臂控制之人機介面，藉此我們找出了手臂移動之角速度與肌電訊號間的映射函數，且能對機械手臂進行變速動作控制。

EMG is a physiological signal generated during muscle contraction. With an appending in implicating the motion intention, the EMG is very suitable to be used for robot control and prosthesis control in an intuitive and directive manner. in this thesis, we analyze the relationship between EMG signal intensity and its corresponding arm movement and velocity, and then construct an EMG-based robot control system with velocity mapping. With this system, the operator can manipulate the robot by moving arm in different speeds. In system implementation, for forearm movement analysis, two electrodes are used to measure EMG signal from biceps and the triceps muscles. Signal analysis is performed in time domain and frequency domain for muscle intensity and activity. Accordingly, a classifier is developed to distinguish forearm movements of different velocities. To achieve real-time control, a human-machine interface is established for signal extraction, signal analysis, velocity mapping, and robot arm control, finally we find out the mapping function between arm motion velocity and EMG signal, then we can control robot arm with different velocities and motions.