以加速度計感應手勢進行機器蛇之遙控

Accelerometer-based Gesture Interface for Remote Control of Snake Robots

黃忠政(Chung-Cheng Huang)

指導教授 ： 周瑞仁

國立臺灣大學/生物資源暨農學院/生物產業機電工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.03029

摘要

本研究設計一個階層式手勢輸入介面，提供使用者利用內建有加速度計的遙控器，以有效而機動的方式來控制機器蛇。階層式手勢輸入介面主要包括手勢辨識及階層式手勢輸入。本研究主要使用加權互相關分析進行辨識，可以辨識控制機器蛇步態所需要的12種手勢和機器蛇形態所需的3種手勢，目前平均辨識率為94.89%，和十個阿拉伯數字0~9及14個英文字母，目前平均辨識率可達93.47%。本研究利用中樞模式產生器模型(CPG)產生連結型機器蛇的步態，除了可用12種手勢來控制已預設好的機器蛇步態，也可以使用這些阿拉伯數字及英文字母來修正機器蛇的行走速度及轉彎角度。其中英文字母代表機器蛇運動參數類別或階層，阿拉伯數字代表參數值。使用者可以用簡單、方便的方式控制機器蛇，只要依參數類別輸入適當的數值即可。本研究所提出的辨識方法除了可辨識上述12種控制機器蛇步態的手勢、3種控制機器蛇形態的手勢，阿拉伯數字與英文字母之外，亦可允許使用者自訂手勢，自訂的手勢只需要輸入30個手勢樣本即可加入訓練，建成手勢資料庫。本研究不只提供機器蛇與操作者一個有用的互動介面，也可應用在一般的消費性電子產品、工業用機器或機電設備的人機介面上。

關鍵字

加速度計 ； 中樞模式產生器模型 ； 手勢介面 ； 機器蛇 ； 加權互相關分析

並列摘要

This study is to design a hierarchical gesture input interface based on accelerometers mounted on a Nintendo Wii Remote Controller to improve the mobility and efficiency for controlling snake robots. This interface consists of two subunits: gesture recognition and hierarchical input approaches. We mainly use weighted cross correlation approaches to recognize user’s gestures including twelve snake motion gestures, three shape gestures, Arabic numerals from “0” to “9” and fourteen English alphabets. The average recognition rate for Snake motion gestures and shape gestures can reach up to 94.89% and for Arabic numerals and English alphabets can reach up to 93.47%. In the control of snake robots using hierarchical input approaches, users not only can directly control predefined snake robot gaits based on central pattern generator (CPG) and shapes of snake robots by using twelve snake motion gestures and three shape gestures respectively, but can also modify the motion speed or direction of snake robots by entering Arabic numerals and English alphabets. Snake motion gestures represent snake robot gaits, shape gestures represent the shapes of snake robots; English alphabets denote parameter types or hierarchical levels and Arabic numerals represent parameter values. Those Snake motion gestures, Arabic numerals and alphabets are enough for the CPG control of snake robots. Besides, the system allows self-defined gestures by providing 30 samples for each self-defined gesture in training phase. The developed human-snake robot interface can be applied to many other human-machine interaction devices such as consumer electronics products, industrial machines or other mechatronic devices.

並列關鍵字

Accelerometers ； Central pattern generator (CPG) ； Gesture interface ； Snake robots ； Weighted cross correlation approach

參考文獻

1. AI motor 1001 Manual. Available at:

2. Blanco, A. G., G. J. Munguia, and P. O. Moreno. 2008. Development of a system for independent ambulation of patients with spinal cord injury using a data-glove and a biped robotic model. Electronics, Robotics and Automotive Mechanics Conference. 461-466. Morelos, Mexico.

3. Eun-Seok, C., B. Won-Chul, C. Sung-Jung, Y. Jing, K. Dong-Yoon, and K. Sang-Ryong. 2005. Beatbox music phone: Gesture-based interactive mobile phone using a tri-axis accelerometer. IEEE International Conference on Industrial Technology. 97-102. Hong Kong, China.

4. Hooper, S. L. Central Pattern Generators. Available at:

5. Jiayang, L., W. Zhen, Z. Lin, J. Wickramasuriya, and V. Vasudevan. 2009. uWave: Accelerometer-based personalized gesture recognition and its applications. IEEE International Conference on Pervasive Computing and Communications. 1-9. Galveston, TX, United States.