Study on Vision Based Object Grasping of Industrial Manipulator

• Main Authors: Chieh-ChunLin, 林潔君
• Other Authors: Ming-Yang Cheng
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/94589257969900874924

碩士 === 國立成功大學 === 電機工程學系 === 103 === In recent years, because of low birth rate and aging population, labor force has become insufficient and labor cost keeps rising up. The manufacturing industries that require a large amount of labor force are severely impacted by these problems. Thus, the topic related to automation has become more and more important. In the meantime, because more and more industries are following the trend of small-volume large-variety production, the flexibility and adaptability of production lines have become much more important than before. In the past, industrial robots that are used for assisting production lines were controlled by teach pendants in order to perform a specific task. The performance and accuracy of industrial robots were highly dependent on the proficiency of technicians. Furthermore, it took a lot of time to set up a production line and it was lack of flexibility. In order to solve the problems mentioned above, the integration of computer vision in automatic production lines is an effective solution. In production lines, robotic arms are frequently used to perform repetitive tasks, such as pick-and-place tasks for packing and placing objects. In the application of the pick-and-place system, the position of objects on a conveyor belt or the locations for placing objects have to be known, in order to place them correctly. Therefore, the topics related to the relationship between object images and object 3D positions and the relationship between object poses and object grasping methods are extremely important. This thesis focuses on the research topic of vision based object grasping of industrial manipulators. This research topic contains object spatial information, object grasping poses, and the eye-hand coordination and frame transformation. This thesis mainly uses the eye-to-hand stereo camera system to retrieve object’s 3D position information by feature matching, and uses shape recognition to give instructions to a robotic arm for picking and placing objects. Experimental results indicate that the vision based automatic system developed in this thesis can successfully complete automatic pick-and-place tasks.