A Flexible Image Processing Framework for Vision-based Navigation Using Monocular Image Sensors

Abstract

On-Orbit Servicing (OOS) encompasses all operations related to servicing satellites and performing other work
\non-orbit, such as reduction of space debris. Servicing satellites includes repairs, refueling, attitude control and
\nother tasks, which may be needed to put a failed satellite back into working condition.
\nA servicing satellite requires accurate position and orientation (pose) information about the target spacecraft.
\nA large quantity of different sensor families is available to accommodate this need. However, when it comes to
\nminimizing mass, space and power required for a sensor system, mostly monocular imaging sensors perform very
\nwell. A disadvantage is- when comparing to LIDAR sensors- that costly computations are needed to process the
\ndata of the sensor.
\nThe method presented in this paper is addressing these problems by aiming to implement three different design
\nprinciples; First: keep the computational burden as low as possible. Second: utilize different algorithms and
\nchoose among them, depending on the situation, to retrieve the most stable results. Third: Stay modular and
\nflexible.
\nThe software is designed primarily for utilization in On-Orbit Servicing tasks, where- for example- a servicer
\nspacecraft approaches an uncooperative client spacecraft, which can not aid in the process in any way as it is
\nassumed to be completely passive. Image processing is used for navigating to the client spacecraft. In this specific
\nscenario, it is vital to obtain accurate distance and bearing information until, in the last few meters, all six degrees
\nof freedom are needed to be known. The smaller the distance between the spacecraft, the more accurate pose
\nestimates are required.
\nThe algorithms used here are tested and optimized on a sophisticated Rendezvous and Docking Simulation facility
\n(European Proximity Operations Simulator- EPOS 2.0) in its second-generation form located at the German
\nSpace Operations Center (GSOC) in Weßling, Germany. This particular simulation environment is real-time capable
\nand provides an interface to test sensor system hardware in closed loop configuration. The results from these
\ntests are summarized in the paper as well.
\nFinally, an outlook on future work is given, with the intention of providing some long-term goals as the paper is
\npresenting a snapshot of ongoing, by far not yet completed work. Moreover, it serves as an overview of additions
\nwhich can improve the presented method further.