Intelligent Depalletizing System Utilizing 3D Image Recognition and Industrial Robot for Camshaft
Department of Mechanical Engineering, School of Aeronautics and Astronautics, Xiang'an Campus, Xiamen University, Xiang'an South Road, Xiang'an District, Xiamen City, Fujian Province, 361102, China
ABSTRACT
Due to labor shortages and the increasing tendency of factories to automate production, the demands for robotic palletizing have increased significantly. This paper introduces an intelligent depalletizing system including an RGB-D depth camera and a six-axis serial industrial robot for handling camshafts. The main functions of system realize material image recognition, handling sequence decision and material handling. Firstly, the core technology of target detection of camshafts is applying instance segmentation algorithm of MASKR-CNN. Then, the average value of center line calculated by mask information involving depth information of each camshaft captured by RGB-D depth camera determines the material grabbing sequence. The smaller the calculated average value, the higher the priority of camshaft grabbing. Subsequently, a self-designed magnetic gripper mounted at the end of the robot realizes the grasping of the center position of the camshaft. In order to reduce the rigid collision during the grabbing process, the gripper adopts the combined design of linear bearing and smooth shaft. Finally, the proposed system achieved a 100% success rate for the grasping results in the case of disorderly placement of camshafts.
Keywords: Robotic Palletizing, Image Recognition, Handling Sequence Decision, Material Handling.
Department of Mechanical Engineering, School of Aeronautics and Astronautics, Xiang'an Campus, Xiamen University, Xiang'an South Road, Xiang'an District, Xiamen City, Fujian Province, 361102, China
ABSTRACT
Due to labor shortages and the increasing tendency of factories to automate production, the demands for robotic palletizing have increased significantly. This paper introduces an intelligent depalletizing system including an RGB-D depth camera and a six-axis serial industrial robot for handling camshafts. The main functions of system realize material image recognition, handling sequence decision and material handling. Firstly, the core technology of target detection of camshafts is applying instance segmentation algorithm of MASKR-CNN. Then, the average value of center line calculated by mask information involving depth information of each camshaft captured by RGB-D depth camera determines the material grabbing sequence. The smaller the calculated average value, the higher the priority of camshaft grabbing. Subsequently, a self-designed magnetic gripper mounted at the end of the robot realizes the grasping of the center position of the camshaft. In order to reduce the rigid collision during the grabbing process, the gripper adopts the combined design of linear bearing and smooth shaft. Finally, the proposed system achieved a 100% success rate for the grasping results in the case of disorderly placement of camshafts.
Keywords: Robotic Palletizing, Image Recognition, Handling Sequence Decision, Material Handling.
