为了实现对空间失效卫星、空间碎片等非合作目标, 尤其是具有自旋运动特性的目标进行在轨服务或者离轨清除, 需要精确完成追踪飞行器与目标飞行器之间的相对姿态测量。首先, 以逆深度参数化表示相机在世界坐标系下的坐标值、高低角、方位角和深度信息, 可以有效解决小视差情况下的单目视觉姿态估计。其次, 建立了相机相对于非合作目标的运动模型和测量模型。最后, 基于单点随机抽样和扩展卡尔曼滤波实现了相机和目标之间的相对运动姿态估计。实验结果表明: 对于三轴稳定目标, 接近过程中姿态测量精度约为0.5°; 对于匀速慢旋目标, 相对角度误差约为3.5%, 平均角速度误差约为0.1°/s。可以满足工程上空间非合作目标相对姿态测量的使用需求。

Purpose: to achieve on-orbit servicing or off-orbit removal of non-cooperative targets in space, such as inactive satellites, space debris, especially slowly rotating targets, accurate measurement of relative attitude between the tracking aircraft and target aircraft is necessary. Method: firstly, used inverse depth parameterization to represent the coordinate value, angular altitude, azimuth angle and depth information of the camera, which could effectively complete the monocular vision attitude estimation under small parallax error; secondly, established a motion model and measurement model of the camera relative to the non-cooperative target; finally, achieved the relative motion attitude estimation between the camera and target based on single candidate RANSAC and EKF. Results: for three-axes stable targets, the attitude measurement accuracy during the approaching process is approximately 0.5°; for slowly rotating targets (constant velocity), the relative angle error is about 3.5%, with an average angular velocity error of approximately 0.1°/s. All these parameters can meet the requirements for relative attitude measurements of engineering spatial non-cooperative targets.