Robot Arm Control with Calculation of Most Suitable Gripping Position

• Main Authors: Shang-Wei Lin, 林上崴
• Other Authors: Po Ting Lin
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/94843527496147578101

碩士 === 中原大學 === 機械工程研究所 === 105 === This paper studies the method of recording and recognizing object shape based on image analysis,which makes the robot arm to select the target object spontaneously,and decide the optimal gripping position. In the part of image analysis, by gray-scale processing,image binarization,image smooth highlights the target object shape and position in the image.Further,the shape characteristic of the object is obtained by the connection domain labeling and the adjacent pixel analysis. Use this characteristic to build database of target object and identify the target object with two-dimensio- nal rotation matrix.Finally, the object as the center of rotation by the Z axis rotate camera to record different angles of the image. Use the absolute position of the centroid obtained by centroid analysis to screen out optimal gripping position and angle. In the part of robot arm control, the rotation angle and speed of each axis is calculated from the coordinates of optimal gripping position by inverse kinematics. Through cartesian coordinate transformation correct the difference between the visual and the robotic arm on different coordinate space.And further commands the arm to move to the specified coordinate to perform the gripping action. This method makes the robot arm to determine automatically the position of the target object placed at random and to choose optimal gripping position by the size parameter of the fixture and the centroid of the target object.The feasibility of this method is proved by actual operation.