Speaker
Description
Space Situational Awareness (SSA) has emerged as a critical priority in the modern space era, driven by the rapid growth of satellite deployments into Low Earth Orbit (LEO). While mega-constellations expand global connectivity, they also intensify orbital congestion and amplify operational risks, challenges further compounded by the increasing involvement of commercial actors and a regulatory framework that has yet to fully adapt.
The research investigates artificial satellites and space debris in LEO, examining the mega-constellation paradigm and its broader implications for the orbital environment and observational astronomy. To mitigate the impact, a deep learning-based solution was developed for the automatic processing of astronomical images, replacing manual frame selection for scientific analyses. At the global scale, this repetitive, non-standardized process is estimated to generate annual economic losses of 150–180 million euros.
Trained on all-sky images from the national MOROI network, the solution is scalable, reconfigurable, and adaptable to individual purposes. The model achieved a precision of 100%, a recall of 91%, and a frame-level accuracy of 95.5%, confirming its readiness for real operational workflows and opening new directions in automated observational data processing.
References:
[1] C. Omat, M. Birlan, D.A. Nedelcu, S. Anghel (2026). Deep learning approach for automated detection of space objects in astronomical imaging. Astronomy and Computing, vol 55, April 2026. DOI: https://doi.org/10.1016/j.ascom.2026.101081
[2] C. Omat, M. Birlan, D.A. Nedelcu, V. Turcu, F. Deleflie, U.E. Botezatu (2025). Re-entry Survivability Analysis of ERS-2 satellite. Romanian Astronomical Journal,vol 35 (1-2), pag 119-134, DOI: https://doi.org/10.59277/roaj.2025.1-2.07
[3] C. Omat, M. Trelia, M. Birlan, D.A. Nedelcu (2024). Analysis of the re-entry phase of Starlink-1353 satellite. Romanian Astronomical Journal, vol 34, pag 75-91, DOI: https://doi.org/10.59277/RoAJ.2024.1-2.05
[4] C. Omat, M. Birlan, D.A. Nedelcu, S. Anghel (2026). Repurposing all-sky camera systems for satellites and debris detection through deep learning, 5th International Conference on Space Situational Awareness (ICSSA), Tres Cantos - Madrid, Spain, 7-9 April 2026
[5] C. Omat, M. Birlan, D.A. Nedelcu (2025). An automatically approach of space objects detection, 9th European Conference on Space Debris, Bonn, Germany, 1-4 April 2025. ESA Proceedings Database: https://conference.sdo.esoc.esa.int/proceedings/sdc9/paper/264/SDC9-paper264.pdf
[6] C. Omat, M. Birlan, D.A. Nedelcu (2025). Real-time Sky Object Detection and Classification using YOLO Algorithm. Prezentare orală la 64th Israel Annual Conference on Aerospace Sciences (IACAS 2025), Haifa, Israel, 20 March 2025. https://www.proceedings.com/79524.html ISBN: 9798331316129