A Study on Image-based Visual Servoing Structure for 6-axis Industrial Manipulator

• Main Authors: Ju-FengWu, 吳如峰
• Other Authors: Ming-Yang Cheng
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/tndd6e

碩士 === 國立成功大學 === 電機工程學系 === 103 === Being the power house of the 3C industry around the world, Taiwan, is facing a serious low birth rate problem. Meanwhile, global demand for 3C products is higher and higher. As result, the problem of lack of worker will be getting worse. It is not surprising that the topic of industrial automation has once again become a hot issue. The 6-axis industrial manipulator is a servomechanism that is commonly used in many industrial applications to perform simple pick-and-place tasks. By equipping with computer vision, a 6-axis industrial manipulator can perform more flexible and complex tasks. Therefore, this thesis employs the image-based visual servoing structure, in which the drives of the servomotors of the 6-axis industrial manipulator are set to torque mode, and a camera mounted on the end effector is used to capture object features. However, when the drives of the servomotors are set to the torque mode, the built-in forward/inverse kinematics software and joint space controllers are no longer available. In order to cope with this difficulty, we develop our own forward/inverse kinematics software and joint space controllers in this thesis. In addition, when implementing the image-based visual servoing structure, one of the most crucial issues is to have accurate depth information. As a result, this thesis is aimed at developing an effective depth estimation algorithm. Experimental results verify the effectiveness of the forward/inverse kinematics software and joint space controllers. Experimental results also indicate that the proposed depth estimation algorithm is indeed feasible.