Research On Pose Measurement Method Of GEO Non-cooperative Target Using Binocular Stereo Cameras

Nowadays, the main aerospace countries are paying more attention to GEOon-orbit servicing(OOS). As a key technology for rendezvous&docking(RvD),pose measurement for GEO non-cooperative target is becoming a research focus. Inthis thesis, a method based on binocular stereo cameras is proposed to measure thepose of a GEO non-cooperative target. Moreover, a semi-physical simulation systemis set up. Based on this system, the proposed method is verified by manyexperiments. The main contributions of this thesis are as follows:Firstly, the properties of typical non-cooperative targets in GEO orbit wereinvestigated. According to the analysis and summarization, the structure withcircular features was chosen as the recognized object. The spacecrafts in GEO orbitwere classified into different types, and the shape characteristics of each type wereanalyzed. From the point of view of pose measurement, some features, such as themain body of a satellite, the bracket of the solar battery board, the communicationantenna, and so on. Considering the measurement range, the target size and the fieldof view, the apogee engine nozzle and the launch vehicle ring are chosen as thecandidates to be recognized for measuring the pose of a GEO non-cooperativetarget.The hand-eye calibration method of the binocular cameras was studied, sincethe transformation relationship between the world frame and the cameras frames arerequired to be known for the pose measurement algorithms. Based on the derivationof the camera calibration equations, the calculation model was simplified andmodified in the practical environment. Then a method based circular feature patternwas used to calibrate the stereo vision system, and validated based on simulationexperiments.A method of pose measurement for GEO non-cooperative target was proposedaiming at the circular features. The key of the method is the combination of thefeature recognition of monocular camera and stereo feature matching of thebinocular cameras. The images of the left and right cameras are processedrespectively, and the desired image features are recognized according to theprocessed results. The recognized features of each camera are then matched basedon the principle of the binocular stereo vision. At last the position of the circlecenter, the normal and the radius of the circle plane are measured. The simulationresults demonstrated the effectiveness of the method.A semi-physical simulation system was established to further verify theproposed pose measurement method. The system is composed of an image-capturing and pose-measuring computer, two image-generating computers and two realcameras. The image-capturing and pose-measuring computer is used to acquireimages on real-time and calculate the position and attitude of the target; theimage-generating computers are used to receive the position and attitude data fromthe image-capturing and pose-measuring computer and driver3D target models toupdate their states; the real cameras are used to capture images and pass them intothe image-capturing and pose-measuring computer. In the experimental study, byanalyzing the errors of the simulation system, the accuracy is improved and theapplicability of the algorithm is confirmed once again.The method proposed in this thesis can measure the pose of GEOnon-cooperative target accurately. The research results have a certain value for thefuture on-orbit servicing in GEO orbit.