| Summary: | 碩士 === 國立勤益科技大學 === 電機工程系 === 107 === In recent years, with the rapid development of technology, automated production tools have been widely used in the manufacturing industry. In the face of the difference in industrial distribution, the demand for robotic arms varies from industry to industry. This paper mainly discusses the integration of image recognition and robotic arm to complete the automatic loading and unloading function on the machine tool. Use monocular vision in the image part, for target detection for the target object. Obtaining the object coordinates in the workspace from the image recognition system, to pass the axial joints of the robot arm through the D-H conversion matrix rule, performing an analysis of mechanical forward and reverse kinematics, the spatial relationship between the Cartesian three-dimensional coordinate system and the joint points of the mechanical arm is obtained. Finally, the communication between the robot arm and the image recognition is established. Combining a fast and robust approach, create a prototype platform that combines automatic image recognition with a robotic arm. This study used a UR5 six-axis robotic arm from Universal Robots in combination with an industrial CMOS camera. Combine computer vision image processing technology Hough transform with Sobel operation, integrated into a system that automatically recognizes objects and accurately clamps them in the machine tool for automatic loading and unloading. The experimental results show that the image recognition system of this study has 1536×2480 resolution and 60 frames per second image information stream to achieve synchronization processing. In a
complex context, objects of different heights can be efficiently and accurately classified. The robot arm can also correctly grasp the identified
object.
|
|---|
