Vision-based Navigation And Tracking Control Of Spacecraft For On-orbit Servicing

The on-orbit servicing is an effectual method to enhance the performance, extends the life, reduce costs and risks of spacecraft, and decrease the number of abandoned spacecraft. The approaching operations to serviced spacecraft such as tracking, rendezvous and docking are basic functionality of servicing spacecraft. Obtaining the relative state information of serviced spacecraft independently and accurately is a precondition for achieving high precision control of approaching operations. Sponsored by the advanced programme of the State Administration of Science, Technology and Industry for National Defence and the CASC advanced found, this paper has theoretically studied the problems such as high-precision relative position and attitude measurement and target tracking control based on monocular vision, facing to the precision control of non-cooperation objects in space. The main innovative research achievements are as follows.The key technology of vision sensors is the arithmetic of calculating relative position and attitude based on target image. Extracting and matching image features using the nature features of the served spacecraft will lead to gross error. In view of this situation, a two-stage iterative algorithm based on inverse projection is given, which contains a depth-of-field estimating and an absolute position calculating. By introducing depth-of-field, this algorithm can eliminate the nonlinear problem caused by pinhole imaging, guarantee the global convergence. In the stage of absolute position calculating, the objective function is minimized and converted into a LMI constraint max/min problem by convex relaxing the non-convex constraints of attitude matrix. And then, it is optimality solved with interior point method. By this way, the algorithm ensures the global convergence, improves the robustness and rapidity. This paper also proves the global convergence of this algorithm. Finally, in the background of on-orbit servicing spacecraft approaching mission, the algorithm is numerically simulated. The results show that in case of feature points extraction errors, this algorithm can also have a faster calculation speed and good convergence.The monocular vision measuring system faces problems like limited output frequency and a lack of depth information in the sight axis. To solve these problems, vision/IMU multi-rate Kalman filter and complementary filter are designed in time-domain and frequency-domain respectively, considering the complementarity of vision and IMU in output frequency and frequency-domain error characteristics. The filtering period of multi-rate Kalman filter is decided by the inertial sensors sampling period. The filter updates time or measurement information at the filtering time depending on if there is vision information or not and then modifies the measurement noise matrix and covariance matrix with the measurement information. The filter can effectively increase the update frequency of navigation filters, and improve the performance of filtering algorithm. A complementary filter is designed based on the different noise characters of vision and IMU in frequency domain and then the stability of the complementary filter is analyzed. The mathematical simulation shows the complementary filter can effectively complement vision and IMU in frequencies, and can be used to estimate the state under quickly changing. Finally, the parametric CRLB model of vision/IMU navigation system is built, and the influence of the vision and IMU parameters on navigation accuracy is analyzed, which provides the theoretical basis for system designed and parameters chose.In the process of approaching, stabilizing the be-serviced spacecraft in the camera’s field of servicing spacecraft with higher precision it’s one of the basic functions of the servicing spacecraft. By introducing the image plane error of target into tracking closed-loop control, a switch gain PD control law for space target tracking is proposed. Using the visual characteristics error in the phase plane directly avoids calibration of camera intrinsic parameters and pose information solver, which simplifies servicing spacecraft system configuration. This mothed improves the control system’s stability and anti-jamming performance in the way of introducing switching function of the algorithm on the PD gain switching. Controller parameter estimation method is proposed and the convergence of the control law is analyzed in Lyapunov theory. Mathematical simulation has been conducted and indicated that the system was able to track the static or complex maneuvering target and the situation considering typical uncertainty factors. The results demonstrate the effectiveness of visual tracking control strategy and robustness for multiple types of uncertainty.With the advantages of low cost and easy to implement, the spacecraft motion simulation system based on the principle of floatation interference is used to simulate the spacecraft translation and attitude dynamic in a small perturbation outer space environment. The scheme of space control technologies verification system is provided and the visual physical simulation experiment of measuring and tracking control system for servicing spacecraft is designed. The calibration of relative navigation vision camera is made. The physical simulation experiment verify the feasibility and effectiveness of single vision position and attitude determination algorithms based on convex relaxation optimized, vision/IMU multi-rate filtering algorithm, and target tracking algorithm based on image information.