The Method Of Non-cooperative Spacecraft Pose Measurement Based On Circular Feature With ToF Camera

In order to extend the lifetime of the spacecraft and improve the reliability of the spacecraft,the on-orbit service system has been started to build.For non-cooperative targets without special identification devices,sensors must be used to measure the target position and posture in order to perform service tasks safely and effectively.Based on the close-range operation task in the orbit service,this dissertation uses the rich-information To F camera as the measurement sensor and uses the circular feature on the non-cooperative target as the key point to study an efficient and reliable method for measuring the relative pose of the non-cooperative spacecraft.Mainly adopting the pose determination method based on the point cloud model,focusing on the research of circlular feature region extraction,space circle fitting and relative calculation and verification method based on circlular feature,the main contents are as follows:To use the circlular feature to solve the relative pose of the target,the circlular feature area must be extracted first.Aiming at the problem,the com plex and inefficient solution algorithm due to the large number of point clouds with depth information and the incomplete data from a single perspective,a point cloud segmentation method using gray-scale image assistance is proposed.Using the particularity of the one-to-one correspondence between the pixel positions of the gray image and the depth image in the To F camera,after the ellipse detection in the gray image,the corresponding position data extraction in the depth image is converted into a point cloud,which point cloud containing a circlular Feature is extracted.Finally,the point cloud of the circular feature area will be obtained after clustering and filtering.Programming experiment with physical shooting data and comparison with point cloud direct segmentation results show that the use of gray-scale assisted point cloud segmentation is faster and more effective.The relative pose calculation based on the circular feature mainly uses the space normal vector and the center of the circle in the circular feature parameter.In order to make full use of the To F camera data and improve the reliability of the results,a method of jointly solving the circular feature parameters using grayscale images and point clouds is proposed.For the normal vector,the end and side characteristics of the circular feature and the ellipse detection results of the gray image are used to solve separately,and finally a reasonable normal vector fusion is performed to obtain a unique solution.For the center of the circle,based on the plane of the standard coordinate system,the depth information of the point cloud and the parameter results of the ellipse detection are jointly solved.Through the experimental results,it is found that the results of each part will be bias ed due to the noise existing in the point cloud data,which verifies the necessity of the result fusion.At the same time,it is found that the grayscale image and the depth image may not be completely synchronized,resulting in deviations in the results.For the problem of using circular feature parameters to quickly solve the relative pose of non-cooperative targets,this dissertation is based on the idea of the model-based pose solution.By establishing a special coordinate system,under the premise that the template is a space circle,the relative pose of the target will be gived by the simple translational rotation relationship of the rigid body.In order to carry out verification experiments without precision equipment,a series of formulas and results verification methods are proposed in this dissertation,and the analysis of the situation where the grayscale image and the depth image are out of synchronization is carried out,and the feasibility of the algorithm is finally proved.