Research On Relative Measurement Algorithm Of Space Rendezvous With A Non-Cooperative Target

Relative states (relative position and attitude) determination between two spacecrafts based on optical measurement theory is one of the most important parts of spacecraft autonomous navigation system, which is also an key technology for rendezvous and docking as well as spacecrafts approaching. Based on the correlative knowledge and technic of digital photogrammetry and orbital mechanics, the optical measurement of relative position and attitude between the spacecraft and the non-cooperated target,which is an important and potential application value field, is investigated in this dissertation.The main achievements and innovations are summarized as following:1). Spacecraft relative orbital mechanics characteristics of non-cooperated rendezvous is deeply studied. The relative motion equation between the chaser and the target is described on the assumption of two-body problem. Flyaround orbit and along-track orbit inspection configurations are proposed according to the equation. Matlab and STK software are used to verify the proposed method. The influence of orbit perturbance, particularly the J2 perturbance, is discussed briefly at last.2). A reasonable assumption is put forward that the target's structure is obtained during the tracked stage, which contributes to a scheme for close-range relative position and orientation estimation for non-cooperated spacecraft rendezvous. Making use of digital image processing technology, the method reduces the observed features on non-cooperated target to detectable image features, such as points, lines, edges and areas in order to solve the problem of lack of known information and finally attains relative position and attitude parameters .Based on the analysis of photogrammetry collinear equation and the imaging mechanism of OpenGL, a simulator is designed to simulate the imaging process of CCD camera, which provides the images to validate the following method. The 3D model of the target is translated into 2D feature points model depends on projection operation and image corner detector, and then a labeling algorithm is brought forward to match the features with the detected corner features correctly. Simulation results show that the method is effective and reliable.3). A fast and globally convergent pose estimation algorithm is proposed on the basis of object-space collinearity equation and error. The problem of photogrammetry space resection is solved by iterated absolute orientation in this algorithm, which operates by firstly obtaining an optimization of the rotation portion based on the least-square problem of revolution of subspace and then estimateing an associated optimization translation.This method, which is global convergent, can strictly guarantee the orthogonality constraint of rotation matrix and automatically provide favorable iterative value. Simulation experiments results suggest that the algorithm can be fast convergent ,which also has advantages of high calculateion on accuracy and strong ability against noise. The proposed method is more suitable for relative position and attitude determination of non-cooperated rendezvous when many unknown factors exist.