ABSTRACT
This chapter presents a pose estimation method for an autonomous approach and docking with an uncooperative known spacecraft. The method includes evaluating the parameters of the relative position and attitude using a computer vision and applying dynamic filtering. Pose determination is based on informative features of images with the use of machine learning, which reduces computational complexity of algorithm implementation. An automatic rendezvous and docking in space is an important problem of modern space technology. The active spacecraft is maneuvering to approach the passive spacecraft for docking. The method of pose estimation refers to a geometric feature learning technique supplemented by the use of a decision tree. Texture features are composed of numerical characteristics computed from local areas covering the entire image.
