Research On Positioning Method Of Vision Guide Assembly For Electronic Manufacturing Robot

In view of the poor assembly accuracy and low efficiency of the current Chinese electronics manufacturing industry,this paper has carried out research on positioning method of visual guidance assembly for electronic manufacturing robot.The assembly object is the flashlight and camera of the mobile phone.This paper combines the development status of robotic vision guided assembly at home and abroad,and focuses on the two aspects of visual positioning algorithm and calibration related to positioning.At the same time,the overall design of electronic manufacturing robot vision guidance assembly system is introduced.The research content and results of this paper are divided into the following parts:Firstly,the hardware composition and installation position,software composition and assembly process of the complete robot vision guidance assembly system are introduced.Among them,the hardware includes robots,industrial cameras,light sources,industrial lenses,high-precision calibration plates and pneumatic control devices;software is divided into interactive layer,functional layer and application layer;assembly process is divided into two parts: camera assembly and flash assembly.Then the calibration study of the robot vision guidance assembly system was carried out.The research was divided into three parts: single camera calibration,two camera calibration and robot hand eye calibration.In this paper,while summarizing related theories,aiming at the problem of large number of cameras,cumbersome calibration process and low calibration accuracy,a method based on affine transformation for robot hand-eye calibration and two-camera calibration is proposed.Then the research of visual localization algorithm is carried out,which is discussed from three aspects: image preprocessing,image segmentation,target recognition and localization.In view of some shortcomings of the existing methods,this paper proposes an improved FT algorithm to achieve the saliency detection of the target.In the aspect of target recognition and localization,an eight-neighbor-based boundary tracking algorithm is proposed to extract the edge contour of the target.At the same time,a least squares optimization algorithm based on noise culling is proposed to complete the edge contour fitting.Finally,through experimental analysis,the visual positioning algorithm's repeat positioning accuracy is less than one pixel,the algorithm time is less than 0.16 s,the total system assembly time is about 30 s,and the assembly success rate is higher than 99.67%.It can be obtained from the experimental results that the research in this paper meets the assembly requirements and has high engineering application value.