Pose determination of large non-cooperative satellite in close range using coordinated cameras

In recent years, some communication satellites lost their ability due to failure of mechanisms to deploy, resulting in large cost. A space robotic system is expected to perform the on-orbit repairing mission. This is a tremendous challenge since the targets are generally non-cooperative, i.e. not any artificial patterns used for the measurement are mounted on the targets. Moreover, these targets are very large. Limited by the FOV (field of view), a monocular camera can not supply enough information of the targets. In this paper, a method based on two coordinated cameras is proposed to determine the pose (position and orientation) of a large non-cooperative target. Firstly, we designed a sensing system used for the non-cooperative measurement according to the investigation of the communication satellites characteristics. Then, a rectangle feature, which is common in the configuration of a satellite, is chosen as the recognized object. Secondly, we proposed that two cameras share the recognition task in a coordinated behavior, i.e. each provides partial image of the rectangle, and the full features are then obtained by fusing their information. Lastly, the corresponding algorithm of pose measurement is addressed. Simulation results of typical cases verify the proposed approach.