针对无法在目标飞行器上布置合作标志器的问题,提出了一种利用矩形太阳能帆板部件测量空间非合作飞行器位姿的方法。该方法首先利用单相机在中远距离(200m-15m)对非合作飞行器成像,然后用基于矩形帆板的视觉特征设计的专用的特征提取算法自动完成矩形特征的选择与提取,最后根据矩形帆板的几何特征约束利用共面四点透视(P4P)算法或门形三线透视(P3L)算法解算目标位姿。与传统视觉位姿测量方法不同,该方法不需要在目标飞行器上布置合作标志器,更不需要人工干预,因此更适用于一般飞行器间的相对位姿测量。实验结果表明,该方法具有较好的鲁棒性,位置误差小于2.3%,姿态误差小于2°,满足对非合作飞行器进行跟踪、接近、绕飞等位姿测量要求。
A vision pose measurement method for non-cooperative space vehicles by using the rectangular solar panel on the vehicles is proposed to solve the problem that the cooperative markers can not be deployed on the target vehicles. The method can be briefly described below: Firstly, the non-cooperative space vehicle is photographed by a single camera in the middle-long distance (200m - 15m), then the rectangular feature is extracted automatically by using the algorithm specially designed according to the visual characteristics of the solar panel, and finally, the pose parameters are calculated by using the high-precision and robust algorithm (the perspective-4-points (P4P) algorithm or the perspective-3-1iness (P3L) algorithm) according to the geometric constraints of the solar panel. Unlike previous pose measurement methods based on vision, the proposed method do not need to deploy cooperative markers on the target, even without human participation, therefore, it is more applicable to pose measurement of general non-cooperative space vehicles. The experimental results indicate that the method is robust to noises, its position error is less than 2.3%, and the attitude error is less than 2°. The method meets the requirements of vehicles' pose estimation during their tracking, approaching and flying.