A Study on Eye in Hand Manipulator with Visual Servo Control

• Main Author: 陳君旻
• Other Authors: 蕭瑛星
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/a9f68d

碩士 === 國立彰化師範大學 === 電機工程學系 === 107 === In this thesis, the visual servo control approaches of eye in hand manipulator based on image features is studied. The camera is mounted on the end-effecter of a manipulator to grab the target image. The obtained errors of image features between the initial positions and the desired positions of a target image are treated as the control command of the manipulator for image visual servo. The main studies include the image processing algorithms for determining the target image feature points and the approaches of using image features to control the manipulator motion. The adopted image processing algorithms such as color segmentation, morphological processing, edge detection, and Hough transform treat the image to get the image feature points of a target. Using the errors between the initial positions and the desired positions of the image feature points of a target, the image Jacobian matrix obtains the camera motion speeds which are transformed into the joint speed command for each joint controller to control the manipulator. The DH convention is used for the coordinate transformation between the joints of the manipulator to obtain the forward kinematics and inverse kinematics. It combine the errors of the target image features from the feedback image to generate the control command for each joint of the manipulator forming a visual servo feedback system, which can control the end-effecter to reach the target position. In this thesis, the simulations and experiments with different number of target image feature points, the image processing algorithms are investigated and the motion paths of visual servo control are examined. These simulation and experimental results show that there are many combinations of initial positions and desired positions of the target image features to demonstrate that the image visual servo is realistic. At the same time, with the target desired position is disturbed so as to understand the robustness of visual servo control for the manipulator. The simulation and experimental results demonstrate the visual servo control is feasible.