Spatial And Non-cooperative Target Pose Estimation Based On Sequential Stereo Observation Images

With the rapid development of space technology,There are more and more space activities generating lots of faulty satellites and space junks.To repair faulty satellites and clean up space junks have become the urgent problems that the space powers must face.However,Most of these targets are non-cooperative targets with no sign of auxiliary measurement which are difficult to track and catch.In this context,the on-orbit service technology is put forward that can realizing the real-time tracking,fixed-point catching and post maintenance of the space targets.It is worth noting that the measurement of spacecraft's relative position and orientation is the base of the on-orbit service and the premise of the above series of operations.Therefore,the method of relative pose measurement for the on-orbit faulty satellites is studied in this article.The main works and invoations are as follows:(1)The research and development of the relative pose measurement of space non-cooperation target are summarized.The stability and real-time performance of binocular vision measurement scheme are determined with a comparative analysis of the main monocular and binocular vision measurement schemes.(2)A corner detection method which is applicable to the space non-cooperation target is proposed.This method introduces a new double diamond template instead of the traditional template by calculating the gray difference of the pixels inside and outside the diamond,This method does not involve the differential operation and have a strong robustness and low computation redundancy.Experiments prove that this method not only improves the efficiency of detection,but also solves the illumination and noise problems that affect corner detection,and provides the beneficial conditions for tracking the subsequent corner points.(3)The motion of space non-cooperative target is difficult to meet the conditions of the traditional KLT(Kanade-Lucas-Tomasi Tracking)optical flow tracking algorithm in practice.Hence the pyramid's optical flow tracking algorithm(PLK)is put forward.When the target motion is not continuous,The PLK algorithm will obtain an unreliable solution of the optical flow vector.The Hessian matrix condition criterion is introduced to improve the PLK algorithm,which makes the tracking corner more accurate and reliable than the traditional algorithm.(4)Based on the two key technologies of corner detection and tracking,a space non-cooperative target pose measurement method based on binocular stereoscopic vision is constructed.In the first step of this method,a preprocess of the sequence image is implemented.Then the key corners of the target image are tracked.For the tracked key corners,the coordinates of the same point are obtained by stereo matching,and the corresponding spatial point coordinates are calculated by the stereo matching.Finally,the relative pose estimation of the target is achieved by the least squares method.The experimental results from the all-digital simulation and semi-physical simulation platform show that the space non-cooperative target orientation error is no more than 12% and the position error is no more than 10%.Compared with the traditional method,the proposed method are more precise and adaptable,which can be applied to the space non-cooperation target in orbit service.