Monocular Vision Pose Estimation For The Final Approaching Stage Of Spacecraft On-orbit Servicing

Spacecraft on-orbit servicing,such as spacecraft maintenance,on-orbit assembly,refueling,de-orbiting and other operations,can reduce the cost of space mission,improve the performance of spacecraft,and extend the life span of spacecraft.The main challenges in this stage are limited camera field of view,incomplete or changing observation characteristics of the target,and high accuracy of pose estimation.According to the problems and challenges faced by spacecraft on-orbit servicing,the paper studies the problem of pose estimation for the last approaching stage of spacecraft on-orbit servicing.The specific research contents are as follows:First of all,for the general modeling method,the orbit motion and attitude motion of spacecraft are separated and described independently,and the coupling effect of the two motions is ignored.In this paper,the dual quaternion is used to model the relative motion of spacecraft and realize the integration of attitude and orbit.It lays the foundation for further research.Secondly,the extended Kalman filter based on multi feature fusion vision measurement model is used to track and estimate the attitude of spacecraft.Non-cooperative target does not have artificial feature marks,but can only be measured by contour features such as solar sail triangle bracket and docking ring.To overcome the drawback of FOV limitation and imaging ambiguity of the camera,a“selfie stick” structure and a self-calibration strategy were implemented,ensuring that part of the contour features could be observed precisely when the two spacecraft approached each other.The dual quaternion is used to establish a unified measurement model for the measurement features(points,lines,circles),and the extended Kalman filter is used to estimate the motion of the spacecraft.The results of the extended Kalman filter are good,but the stability time is long,and some estimation errors are slightly large.Thirdly,in order to improve the performance of relative pose tracking and estimation in the dynamic state of the filter,a strong tracking filter is introduced to track and estimate the spacecraft.In view of the selection of weakening factors and the adaptive calculation of multiple suboptimal fading matrix in the strong tracking filter,this paper proposes an improved strong tracking Kalman filter with the characteristics of fuzzy adaptive.In this method,a fuzzy logic adaptive controller(FLAC)is used to adjust the weakening factor dynamically,so as to adjust the multiple suboptimal fading matrix adaptively "online",so as to improve the rapidity and stability of the system.Compared with the extended Kalman filter,the estimation result of the strong tracking Kalman filter has a significant improvement in rapidity and stability.Compared with the strong tracking Kalman filter,the estimation result of the improved strong tracking Kalman filter is more stable.Finally,in order to verify the effectiveness of the algorithm in the practical engineering application,a hardware-in-the-loop simulation verification system of monocular vision relative navigation system is built.Taking the spacecraft model as the target spacecraft,the relative orbit motion and attitude motion of the spacecraft are simulated by the guideway with translation and rotation functions.A monocular camera is used to photograph the moving spacecraft model.The monocular vision measurement software in MATLAB simulation environment is used to process the image,extract the features and estimate the relative position.The hardware-in-the-loop simulation results verify the accuracy and effectiveness of the improved strong tracking filter for non-cooperative spacecraft pose estimation.