Thomas Delaite, Onera; Jocelyn Couetdic, Onera; Eric Glemet, Onera; Frederic Cassaing, Onera
Keywords: Image processing, SSA, Optical detection
Abstract:
Space Situational Awareness (SSA) becomes more critical as space gets more and more congested. Following the successful development of the GRAVES radar for LEO satellites, Onera is considering an optical ground station to track Resident Space Objects (RSOs), targeted primarily for altitudes above 2,000 km. Two main approaches are reported in the literature: streak detection in single long exposure frames and Track-before-Detect (TBD) algorithms applied to image stacks. The recent availability of efficient components like sCMOS cameras and GPUs makes on-the-fly TBD a realistic and affordable solution.
Our objective is to specify, optimize, integrate, and validate the most efficient hardware and software components for a prototype ground station, ensuring optimal performance and reliability of the system. The performance of the system needs to be characterized to enable future orbit determination.
The first step of the workflow involves a detection algorithm based on a TBD method, which belongs to a class of tracking methods operating directly on raw temporal series of images. The proposed adapted version, inspired by exoplanet detection research,simultaneously estimates the position and velocity of faint satellites with high accuracy. While this approach typically demands increased computational power for real-time processing of incoming image data, the high acquisition frequency leads to simple motion and shape of the objects in the image, resulting in a fast and efficient algorithm. Moreover, the resulting algorithm is highly parallelizable, making its implementation on GPUs significantly reduce the computation time.
The second step involves the accurate projection of the true state of an observed object into the measurement space. Achieving this requires precise temporal stamping of the data, which must account for the camera’s reading mode. Plate solving is used to compute the projection function that maps pixels to the sky. Conventional methods from astronomy need to be analyzed and adapted to meet SSA requirements.
Finally, an extensive measurement campaign was performed to explore the performance of the system with on-sky data. The system accuracy is analyzed in terms of residuals between our measurements and the projection of calibration satellites’ states. Therefore, the analysis not only measures our detection accuracy, but the overall capacity of the system to track RSOs. It demonstrates sub-arcsecond precision in position and under the arcsecond per second in velocity. An emphasis is put on obtaining residuals over the full range of observations parameters. This allowed to highlight, for example, an increase in the error at low altitude. In addition, the system’s empirical performance in low SNR conditions are investigated on faint objects taking advantage of the satellites’ passage into eclipse. This characterization of a reliable error model is the key to future orbit determination.
This work is supported by PhD Grant DGA/AID 2021305.
Date of Conference: September 19-22, 2023
Track: SDA Systems & Instrumentation