GYSELE : an attitude estimation and non-stellar object tracking system
- Paper ID
89337
- DOI
- author
- company
Centre National d'Etudes Spatiales (CNES)
- country
France
- year
2024
- abstract
GYSELE (GYro-Stellar Enhanced Localization Eyeing in space) is an in-house payload demonstration project led by CNES (French Space Agency). More than just an instrument or a sensor, it is a complete estimation and tracking system, including hardware (gyroscope and camera) and software to be hosted on an on-board computer, capable of both delivering accurate satellite attitude estimate based on star measurements and identifying then tracking artificial objects crossing the field of view of the camera (in the range of magnitude of detected stars). GYSELE offers interesting perspectives for applications such as space-based surveillance and tracking or interplanetary navigation (or even close vision-based navigation around small celestial bodies), as well as valuable support for formation flying, while simultaneously ensuring accurate attitude estimation. This principle is particularly well-suited for the restitution of trajectories of non-stellar objects, as it allows overcoming issues of inter-head error that occur when attitude estimation and measurement of the direction of the object of interest are performed by two different optical sensors. GYSELE development benefited from ESTADIUS heritage, a flight proven day and night 3-axis attitude star sensor based on gyro-stellar estimation developed by CNES for its stratospheric balloon gondolas applications. Indeed, GYSELE adopts ESTADIUS algorithmic principles for the attitude estimation part based on a 15-state Kalman filter hybridizing high-rate (5 to 30Hz) gyroscope measurements and low-rate (1Hz) multiple star direction measurements obtained from images acquired by the camera, and some of its software architectural pattern and features draw on that of ESTADIUS (module task sharing, integration time depending on angular rate, star selection, acquisition and tracking modes). GYSELE algorithms are developed using Matlab/Simulink\textregistered, and executable codes are automatically generated by the Matlab/Simulink\textregistered{ }C code generator. Subsequently, these codes are integrated into the embedded software following a generic architecture adopted by CNES (LVCUGEN standing for Generic Payload Flight Software). The paper will present hardware and software architecture of GYSELE, the classical issues to be addressed in the development (timing synchronization, sufficient signal-to-noise ratio detection ensuring permanent tracking, algorithm complexity and tuning, level of autonomy and FDIR, need for realistic simulation) and in the validation process relying on both high-fidelity simulations (Matlab/Simulink\textregistered), tests performed on hybrid benches (combining both numerical models and actual hardware) or final night-sky tests with every hardware parts in the loop. Performance results are presented for a mid-cost configuration (\begin{math}0.01^{\circ}/{\sqrt{h}}\end{math} class gyro and 3D+ visible camera built with CMV4000 detector and f/3.5 25mm focal Optics).