三軸加速規型復健用途資料手套之開發

Abstract

資料手套(data glove)是一個外觀類似手套的人機互動輸入裝置，主要係用以擷取裝置在手指(或手掌)上之許多個感測器的物理資訊 (例如：手指彎曲角)後，傳送至電腦，以進一步做必要之分析及處理；對於手部受傷的術後病患而言，若能將此資料手套搭配互動式的虛擬實境系統作復健治療，則患者將能樂在其中，效果倍增。 市售的資料手套所使用的感測器種類繁多，而這些感測器的特性通常有體積過大(有的甚至需搭配機械結構，不適用於孩童)、線性程度不良、裝置易於裂化等現象，有鑒於此，本研究採用兼具體積小、重量輕等特色之三軸加速規(accelerometer)為主軸來研發一新型的資料手套，以期能取代現有資料手套所使用之感測器。 本論文針對加速規應用技術提出一新的方法論，引用牛頓力學原理，並藉由數學推導方式而得到手指關節之彎曲角度。 有關實驗方法，本研究採用2個ADXL327加速規貼附於小型機械臂上以模擬人的手指關節，並採用16 bits之MSP430F1612微控制器為其系統核心，使其可執行即時之A/D轉換，並可依序執行即時之加速度、傾斜角以及彎曲角之數學計算，再將以上計算結果經由一I/O介面裝置而傳送至個人電腦，並透過LabVIEW軟體顯示其數據及波形。此實驗裝置後經數據實測結果，在一般使用情況下，其角度誤差值範圍皆可在3度以內，此相較於目前治療師所常使用之量測精度為5度的手指關節量角器而言，本研究已能符合需求。

A “data glove” is a glove-like input device for human-computer interaction, It's commonly used to acquire the physical information (e.g. bending angle of the finger) from several sensors attached on the fingers (or palm) for further data analysis and process. For previous study, the virtual reality can promote patients’ will to receive hand rehabilitation program and provide more efficiency than conventional treatments. There are several types of data glove, all of which are different in style and design, However, some disadvantages such as large size (some are not adequate for children), nonlinearity, frangible, etc. limited the application in treatments. Therefore, in this research, we try to apply a small volume and light weight 3-axis accelerometer to develop a novel data glove for hand rehabilitation without the drawbacks of conventional data gloves. This thesis proposed a new methodology for application technology of accelerometer. It use Newton's Mechanics and by the method of mathematical derivation to get the bending angle of the finger joint. Two ADXL327 accelerometers attached on a small mechanical arm were used to simulate the human finger joints, and a 16 bits MSP430F1612 microcontroller was chosen as the system core, it performs real-time A/D conversion, real-time calculation of the acceleration, tilt angle and bending angle in sequence. The calculated joint angle was transmitted to the PC through a I/O device and display its data and waveform in LabVIEW environment. The results showed the error range of the angle is within 3 degrees in general case. Compared to the joint goniometer with accuracy of 5 degrees for measuring the finger bending angle, this research has been able to meet the demand.