Initial Relative Orbit Determination for Camera-based Navigation in Rendezvous with Uncooperative Target

The newest concepts of rendezvous missions involving on-orbit servicing tasks and proximity operations have highlighted the need for alternative strategies for the Target's orbit determination. While a cooperative Target is usually a spacecraft equipped with Global Navigation Satellite System (GNSS) sensors and Inertial Measurement Units, and is therefore capable of communicating its state to the Chaser or to the Ground Segment, this does not hold in the case of a non-cooperative Target, like a decommissioned or malfunctioning satellite, an asteroid or a space debris. Angles-only navigation offers a relatively simple and low cost solution to this problem, exploiting a Chaser-mounted optical camera which tracks the Target motion along the entire orbit. The main disadvantage of angles-only navigation is represented by the complex determination of the inter-satellite distance, which is unobservable if linear motion models are employed. This Thesis presents a method for Initial Relative Orbit Determination (IROD) that leverages Line-of-Sight observations and the non-linearities in the relative motion dynamics to retrieve the initial relative position between the two spacecraft, which can be subsequently used to initialize a real-time navigation filter. Furthermore, given the space hardware requirements and constraints, this method must be efficient and computationally light enough to run on mission-feasible hardware. In the long term, this could enable the design of fully autonomous spacecraft capable of accomplishing delicate tasks like orbit determination indipendently, without necessarily relying on data provided a priori or uploaded from ground. After a brief introduction on rendezvous and proximity operations, with a focus on the contribution of DLR to the history of rendezvous and on-orbit servicing missions, the Thesis will present the theoretical fundamentals of relative motion dynamics, concentrating on analytical relative motion models suited for on-board application. Subsequently, angles-only navigation will be covered thoroughly, with a deeper insight on the Initital Relative Orbit Determination scope, challenges and possible solutions. Chapter 4 will evaluate the accuracy of analytical relative motion models, and will introduce a mathematical formulation which models the nonlinearities associated to the curvature of the orbital path. In Chapter 5, the implemented IROD algorithm will be presented and extensively discussed, covering its fundamentals, strengths, assumptions and the optimizations that have been performed. Finally, Chapter 6 will offer a detailed evaluation of the algorithm performance, by highlighting its robustness in presence of noise and visibility constraints, presenting tests based on Monte-Carlo analyses, assessing its behaviour with real data from the PRISMA mission of DLR, and finally evaluating its runtime performance on a real on-board computer.