Research On Key Technologies Of Robot Assembly Based On Vision And Force Perceptio

Assembly robots are widely used in various fields of industrial production,and play an important role in improving assembly quality,saving labor costs,and reducing labor intensity.In the robot assembly system,due to the influence of the error of the robot and the error of the assembly platform,it is difficult for the robot to successfully complete the assembly task.In order to calibrate and correct the error in the assembly system,the error calibration is carried out based on the combination of vision and force perception to improve the calibration accuracy,and the automatic calibration process is realized through virtual simulation assembly.This paper mainly focuses on four key technologies: error calibration,visual positioning,force control and simulation assembly.The following researches are carried out:Aiming at the error in the robot assembly system,an automatic error calibration scheme based on the combination of vision,force perception and simulation technology is proposed.In order to improve the applicability of the calibration scheme,a general calibration block is designed that is not limited by specific assembly features.On this basis,using the plane motion principle of the rigid body as the mathematical model,according to the simulation coordinates and the actual coordinates of the calibration block,the error calibration of the robot assembly system is completed.In terms of visual localization,a visual localization strategy based on the combination of template matching and contour detection is proposed.The approximate position of the calibration block is obtained by using Scale Invariant Feature Transform(SIFT),which is robust to changes in scale,rotation,and illumination.On this basis,the Arc Adjacency Matrix-Based Ellipse Detection(AAMED)is used to detect the calibration holes on the calibration block,so as to realize the rough positioning of the calibration holes.In terms of force control,an ensemble model based on the fusion of e Xtreme Gradient Boosting(XGBoost)and Decision Tree was trained to predict the adjustment direction in the process of force control.The adjustment step prediction method on the change of subtraction function curve is proposed,and finally the accurate positioning of calibration hole is realized.In order to realize the automatic calibration technology based on vision and force perception,in the simulation assembly environment of Visual Components,a method is proposed to indirectly realize the control of PLC,industrial camera,force sensor and other equipment by the robot program based on the output signal of the robot.The data communication between the two parts realizes the automatic calibration of the error of the assembly system.Through the integration of visual positioning,force control and other related algorithms,the upper computer software is developed to facilitate automatic error calibration and correction.Finally,the effectiveness of the scheme is verified through the accuracy measurement experiment and the assembly experiment of the product "cabin".