| Summary: | Pose (position and attitude) measurement of a space non-cooperative target is very important for on-orbital servicing tasks, including malfunctioning target repairing, space debris removal, and so on. However, such targets are generally non-cooperative, i.e., no markers are mounted on it and there is no prior knowledge. Therefore, the identification and measurement of a non-cooperative target is very challenging. In this paper, we propose an efficient method to recognize the natural objects with circular or near-circular shapes on the target, reducing the computation load and improving accuracy. First, the geometry properties of a practical non-cooperative target are analyzed. A stereo vision system is correspondingly designed to measure the relative pose of the target. Second, the error sources and time-consuming factors of the traditional method are analyzed. Then, a solution concept is proposed. Third, the efficient method is detailed to solve the geometry equation and determine the pose information, reducing the calculation complexity and increasing the accuracy. The image pre-processing and target detecting algorithms are realized on FPGAs, further accelerating the calculation speed. Finally, we develop an experiment system and verify the proposed method through practical experiments. The experiment system is composed of satellite mockup, binocular camera, and high precision laser tracker. The experiment results show that the proposed method has high accuracy and efficiency.
|
|---|
