Research On Vision Positioning System Of Robot Assembly Based On Multi Camera

Visual positioning technology is one of the key technologies used in robot assembly system.At present,monocular,binocular or stereo cameras are mainly used in visual positioning.However,these devices have some problems,such as poor anti-occlusion ability,positioning accuracy is difficult to meet the requirements of assembly operation.How to improve the anti-occlusion ability and positioning accuracy of visual positioning in assembly operation system is an urgent problem to be solved.Therefore,this paper proposes a visual positioning system for assembly based on multiple cameras.(1)First of all,we introduce the overall framework of the assembly operation system,and analyze the current visual positioning scheme.In order to improve the positioning accuracy and environmental adaptability of the traditional binocular vision system,this paper adds cameras on the basis of the traditional binocular positioning,forms a multi-camera positioning system,and builds a multi-camera positioning system platform.(2)Secondly,aiming at the problem of low accuracy of 3D reconstruction of feature points,a multi-camera 3D reconstruction method based on camera weight is proposed,which is used to locate the target points in the laboratory environment.The error in the process of 3D reconstruction is analyzed.According to the different contribution of each camera in the reconstruction process,the calculation method of camera structure weight parameters is proposed.(3)Then,aiming at the problem of noise interference in camera positioning,in the process of data denoising,considering the spatial structure characteristics of the measured object,according to the spatial constraint relationship of feature points,a 3D positioning data denoising method based on spatial constraint is proposed,which improves the spatial positioning accuracy of multiple cameras.(4)Finally,the results of the previous research are verified by experiments,and the positioning accuracy of the visual positioning system is verified by using fixed position markers.The experimental results show that the measurement accuracy of the proposed algorithm is significantly improved compared with the traditional least square algorithm.After using the noise processing method based on spatial constraints,the measurement accuracy is further improved.Finally,in the case of obtaining the pose of the target workpiece by the multi-camera positioning system,the screw piercing experiment is completed by using the 6-DOF robot,and the experiment proves the correctness and effectiveness of the method.