Vision Measurement Technology For Automatic Assembly Of Segments

Shield machine is a large-scale special engineering machinery equipment that integrates machinery,photoelectricity,hydraulic pressure,electric control and other disciplines.It has a wide range of applications in tunnel engineering.With the increase in the size of the city and the rapid development of rail transit,shield machines play an important role in the field of tunnel construction.In the tunnel shield construction,the assembly of segments has always been a very important part of the construction.It can not only maintain the structural form of the tunnel,prevent the tunnel from settling,but also prevent groundwater leakage and provide a safer construction space.At present,most of the domestic pipe segment assembly requires manual operation of the pipe segment assembly machine for transportation and assembly.The speed and accuracy of the pipe segment assembly depend on the proficiency of the workers.In the entire process of assembling the segments,problems such as broken segments and uneven gaps between segments and segments will occur,which will affect the speed and quality of tunnel construction.This article takes the segment assembling machine as the research object,and uses visual measurement technology to measure the posture of the segment to be assembled.Aiming at the problem of different size of the tube pieces,a scheme of fixing the target on the tube pieces to indirectly obtain the position and posture of the tube pieces is adopted.The main research work of this article is as follows:1)Firstly,the camera imaging model and calibration technology are analyzed theoretically,and the binocular camera is calibrated by Zhang's calibration method.The analysis results show that the camera parameters obtained by the calibration experiment are consistent with the actual.A scheme for indirectly measuring the position and posture of the segment using a cooperative target is designed,and the principle of position and posture measurement of the segment is described.Combined with the research status of point feature pose analysis,the number of feature points required for the target is determined.The effect of target size on measurement accuracy is analyzed,and an image algorithm process for real-time target recognition is designed and four feature points of the target are accurately extracted to prepare for pose calculation.2)Secondly,the binocular vision and n-point perspective problems are combined to solve the target pose.For the theoretical model of binocular vision measurement,the measurement accuracy of the triangulation method and the ray intersection method is analyzed from theory and experiment.The experiment shows that the ray intersection method has higher measurement accuracy;for the current PnP algorithm,different algorithms are compared in the experiment at the measurement accuracy when the feature point is 4,the rotation matrix solving algorithm with higher accuracy is selected.3)Finally,a six-degree-of-freedom visual posture measuring device was built to verify the resolution accuracy of the visual algorithm,and the absolute error of the visual measurement posture was verified with the help of a high-precision checkerboard.The results show that the fusion of binocular vision and n-point perspective problems can improve the accuracy of position and pose solution,and the relative error of translation does not exceed 1mm,the absolute translation error of x-axis and y-axis is less than 0.6mm,and the relative and absolute error of rotation angle is less than 1 degree.