發展遙控機器人之影像搖桿技術

Development of Image Joystick Technique for Telerobotic Systems.

歐新正(Shing-Cheng Ou)

指導教授 ： 何明果

大同大學/工程學院/機械工程學系所/碩士(2008年)

摘要

本文描述透過網路攝影機接收操作者肢體上的標籤影像，並將此標籤色塊視為影像搖桿，用以操控遠端機器人。
視覺搖桿技術為利用網路攝影機擷取使用者的顏色標籤，此顏色標籤分別貼在頭部戴上的帽子中心與前緣及手部末端。當肢體擺動時（如頭部轉動、手腕運動等），色塊標籤也隨之移動；
以下為標籤色塊搜尋與計算機器人馬達轉動量之過程:
(1) 應用不同的標籤色塊的顏色特徵向量，可透過網路攝影機搜尋符合此特徵向量的像素。
(2) 透過彩色系統的形態學將目標色塊萃取出來，再計算出色塊形心位置。
(3) 透過逆運動學與座標轉換與平移即可評估出相對應的肩膀及手肘轉動角度與頭部左右及上下的擺動範圍。
(4) 利用程式估算機器人相對應之馬達轉動角度，透過網路進行遠端控制。
機器人將完全模擬遠端使用者的動作，使之成為一以影像搖桿控制之網路機器人。
影像追蹤系統特徵如下：
(1) 適用於複雜環境
(2) 可求取多個目標色塊
(3) 誤差在2個像素以內
(4) 機器人反應時間僅100毫秒。

關鍵字

遙控機器人 ； 影像搖桿

並列摘要

This article describes the tag image of operator’s limbs which is received through webcam, and takes the tag images as the image joystick to control the telerobotic.

The technique of image joystick comes from the retrieving tag images through webcam. The tag images should be stuck on the center and edge of the operator’s hat and the end of his hand. When the limbs liberate (such as the rotation of the head or motion of the hand) and the tag image librates as well.
Following is the procedure of searching tag images and calculating the angles of the robot’s motors.
(1) Applying the different eigenvector of tag images, it can search the pixels which correspond with these environments through webcam.
(2) To extract the target tag image by color morphology, then calculate the center point of the tag image.
(3) We can evaluate the corresponding angle of operator’s shoulder and elbows rotation and the liberation range of his head by inverse dynamics and coordinates translation and shift.
(4) Using program calculate the corresponding angle of motor, we control the telerobotics by network.

Then the telerobotics will modify the action of the remote operator, so it is able to be a telerobotics controlled by image joystick.

The features of this image joystick used image tracing system as follow:
(1) Applying the complex environment.
(2) Approaching the multi-objects.
(3) Tracking error is under 2 pixels.
(4) Telerobotic reaction is only 100 ms.

並列關鍵字

image joystick ； telerobotic

參考文獻

[2] Sandro Mussa-Ivaldi, 2000. “Neural engineering: Real brains for real robots.” Nature, v 408, p305-306.

[4] Craig, J. J. “Introduction to Robotics : Mechanics and Control, Third Edition” 。

[7] Ivan Petrovic, Misel Brezak，“Machine Vision Based Control of the Ball and Beam” ，2002 IEEE。

[8] M-C. Tsai, K-Y. Chen, M-Y. Chen and K.C. Lin，”Implementation of a real-time moving object tracking system using visual servoing ” ，March 27,2003。

[11] 蔡宗成，胡正鈺，陳明周，”遠端網路代理機器人之設計與應用”，2001。