Research On Axis Coding Recognition And Axis Location Vision Technology Of Traction Shaft Machining Subsystem

In order to improve production efficiency and reduce labor costs,enterprises upgrade the production line and replace manual loading and unloading by robots.However,in the process of production,it is found that the position of the traction shaft in the machine tool will change due to the environment,which leads to the robot can not grasp the traction shaft accurately,and sometimes causes damage to the production line and machine tool in the process of loading and unloading,which increases the production cost of the enterprise.Aiming at the above problems,in order to increase the intellectualization of the system and avoid the occurrence of robot error grabbing,this paper starts from the following two aspects to solve the existing problems of the traction axis processing subsystem,mainly from the research of the axis coding recognition and axis positioning vision technology of the traction axis processing subsystem.The main contents are as follows:The subsystem of axle processing is briefly introduced.The problems existing in the existing system are described in detail,and the solution of using the coding recognition of the traction shaft to coordinate with the final positioning of the traction shaft is proposed.The visual system is studied in detail,including the selection of camera,light source and lens.For the reformed system,the correct recognition of the axis coding is the basis for the completion of axis positioning.The position information of the axis itself(the distance between the center of the traction axis and the end of the axis)needed in the process of axis positioning corresponds to the coding one by one.Firstly,the feasibility of the recognition method of the traction axis coding is studied.This paper uses the improved YOLO algorithm to identify the traction axis coding.Based on the location principle of the YOLO algorithm,the backbone network structure of the improved algorithm is constructed.The algorithm is trained in the format required by the network and tested in batches in the laboratory environment.The test results show that the recognition accuracy of the algorithm is high.The position of the robot grasping the traction axis is the center of the traction axis.Because of its uncontrollable movement in the axial direction(horizontal direction),the robot can not grasp it accurately.So this paper uses visual method to locate the traction axis.Because the scene environment is complex,it is impossible to directly obtain the position of the traction axis center in the reference coordinate system by vision.So this paper first uses vision to obtain the position of the end of the traction axis in the reference coordinate system,and then determines the position of the traction axis center in the reference coordinate system,i.e.the grasping point of the robot,according to the information obtained from the coding recognition in the previous step.Finally,the vision system is experimented,and the traction axis with different coding is selected for the recognition experiment.The experiment shows that the improved YOLO algorithm has high recognition accuracy for the axis coding and meets the actual needs of the scene.The traction axis with the largest and the smallest axis diameter is selected for the visual positioning experiment respectively.The experiment shows that the position of the traction axis based on visual positioning is accurate.Indeed,meet the actual work requirements.The key problems of axle positioning of the system have been solved,and the expected results have been achieved.