Yohann Felicite, ISAE-SUPAERO; Kanthanakorn Noysena, National Astronomical Research Institure of Thailand (NARIT)
https://doi.org/10.64861/DRZH5232
Keywords: Space Situational Awareness, Solar Glints, Embedded Detection, Pre-Launch Development, IMX500 Sensor, Physics-Based Simulation, CubeSat
Abstract:
We present a simulation-driven methodology for the pre-launch development of embedded solar glint detection algorithms, addressing a fundamental constraint in Space Situational Awareness (SSA): no orbital training data exist prior to satellite launch. While ground-based SSA systems require extensive observational datasets, we demonstrate that physics-based simulation can substitute for on-orbit data during algorithm development, thereby reducing technical risk for CubeSat SSA missions.
Our simulator integrates five modular components: GalSim point spread function modeling, the Gaia DR3 stellar catalog with zodiacal light background, the IMX500 sensor noise chain, bidirectional reflectance distribution function (BRDF) spacecraft material models, and SGP4 orbital propagation with Sun-satellite-observer geometry. BRDF reflectance models are validated against laboratory measurements, and sensor parameters are characterized through hardware testing. Parameter uncertainties, particularly Attitude Determination and Control System (ADCS) jitter, which cannot be characterized before flight, are explicitly bounded.
The detection algorithm targets a 3U CubeSat in sun-synchronous Low Earth Orbit (LEO) at 500 km, using a Sony IMX500 sensor and a Raspberry Pi Zero 2W processor. Performance validation demonstrates 14 fps throughput and an F1-score of 88.8% for LEO debris detection. Detection is material-dependent: high-albedo debris achieves robust recall, while low-albedo surfaces (e.g., solar panels) produce insufficient flux. Detection performance in GEO/MEO regimes shows limited recall due to velocity-based filtering optimized for LEO debris. This work establishes a replicable template for future embedded SSA missions at National Astronomical Research Institute of Thailand (NARIT).
Date of Conference: September 15-18, 2026
Track: Space-Based Assets