Optimisation of Star Trackers for Simultaneous Resident Space Object Detection and Attitude Determination

Abstract

Safe and sustained access to space is increasingly being threatened by the growing number of satellite launches and large amounts of space debris. Having the ability to detect and track these Resident Space Objects (RSOs) is crucial to predict and avoid possible future conjunction events. Star trackers, which are common sensors on board satellites, are increasingly being considered for the task of RSO detection, given their prevalence and the benefits they may offer over existing sensors. In this thesis, the CROSS star tracker is evaluated for the simultaneous functions of attitude determination and RSO detection. Using high-fidelity simulated image datasets, a pipeline to perform attitude determination and RSO detection is developed and tested to prove the feasibility of such a system. Optimisation methods including grid search and Bayesian optimisation are also used to improve the system performance, including the attitude accuracy, RSO detectability, and sensor exposure time. The best performing configuration had an average attitude error of 22.5±26.7" and detected 286 RSOs in a single orbit. This improves on prior work in which using a star tracker for RSO detection resulted in unacceptable attitude determination performance or the impact on attitude determination performance was ignored altogether. Using the simulated datasets, the system's sensitivity to intrinsic and extrinsic parameters is also explored. Finally, the CROSS star tracker is further developed in order to ready it for in-orbit demonstration, allowing for the future testing of the proposed system. In all, this work supports the use of star trackers for enhanced Space Situational Awareness (SSA). With future testing to demonstrate their ability to perform Initial Orbit Determination on the detected RSOs, star trackers may be key to expanding our SSA capabilities.