Research On Visual Positioning Control System Of Industrial Robot

With the development and maturity of information technology and the proposal of the concept of Industry 4.0,the demand for improving production efficiency in the manufacturing field is becoming more and more urgent,and the demand for intelligent production is also getting higher and higher.It has gradually become the mainstream trend in the development of industrial automation manufacturing technology to replace human beings with industrial robots for production and manufacturing.However,the vision system matched with industrial cameras can not only help industrial robots to cope with complex processing and manufacturing environments flexibly,but also change the production and processing modes according to actual production requirements.Therefore,the development of visual positioning control technology for industrial robots is an important method to improve the production efficiency of industrial robots.In this paper,Yaskawa robot is taken as the main application carrier of this design.Industrial cameras and panel light sources are assembled at a fixed viewpoint to collect the working plane.At the same time,Beaver vision processor is combined as the upper mechanism to form a complete industrial robot vision control system and the following aspects are deeply studied:1.In this paper,the target image preprocessing method is designed specifically,Gaussian filter is selected to eliminate image edge noise,and top cap transformation is used to deal with the edge blur problem caused by inconsistent photosensitivity of workpiece image contour based on the problem that the target image is easily affected by illumination changes and shadow occlusion in unstructured environment.At the same time,Sobel operator is used to complete image contour extraction and edge matching template establishment,and the sum of normalized edge gradient vector products is used as matching similarity measurement to enhance the anti-illumination and anti-occlusion capability of the matching algorithm.In the process of traversal matching,image and template pyramid are constructed for hierarchical search,and rotation transformation is introduced into template search to solve the problem that workpieces cannot be matched due to rotation and improve the recognition rate of target images.2.A direct calibration method for image coordinate system and robot base coordinate system is proposed which aims at the calibration accuracy problem of coordinate coupling between robot and camera.This method analyzes the problems of large working plane,obvious camera distortion and complex acquisition environment that are often encountered in vision calibration under fixed viewpoints,and eliminates camera distortion by interpolating and subdividing the acquired calibration points to construct a direct conversion relationship between image coordinate system and robot base coordinate system.This method avoids the error problem caused by over-reliance on camera pose correction,distortion correction and other algorithms in the traditional algorithm,and realizes high efficiency and high precision hand-eye calibration with an average error of less than 1 mm.3.In this paper,a two-stage hybrid heuristic algorithm is proposed based on the path optimization strategy of each stroke when simulating robot writing Chinese characters which aims at the C-TSP problem that industrial robots need to plan multiple target paths during processing.The multi-objective path problem is transformed into a single-path TSP problem by improving the heuristic function,and the ant colony algorithm is introduced to solve it.In order to solve the problem of slow convergence speed in the early stage of ant colony algorithm,this paper uses potential field method to improve the initial average distribution of pheromones in the algorithm,and introduces reward term function to optimize the update strategy of pheromones.At the same time,the local path search method is designed by introducing random terms to avoid the algorithm falling into the misunderstanding of local optimization.Finally,this paper perfects the system software design based on each part of the function,and designs experiments of each part of the function to test the overall performance of the system.The experimental results show that all the performance indexes of the system can meet the actual use requirements of industrial robots.Besides,the system can complete the precise control of the industrial robot according to the production and processing tasks in the actual production environment.