Jeremy Correa, Katalyst Space Technologies; William Oldroyd, Katalyst Space Technologies; Jeff Uyekawa, Katalyst Space Technologies
Keywords: SDA, Rendering, GPU, Optics, Imagery, SNR
Abstract:
With the rapid increase in satellite missions by both government and commercial entities, advanced operational data processing and simulation capabilities have become critical to maintain effective space domain awareness (SDA). A central challenge in SDA is not only detecting these objects at ranges spanning LEO, GEO, and into the cislunar domain but also accurately characterizing them, especially when they are dim, non-cooperative, or potentially adversarial. When an object is first detected, little may be known about its nature; therefore, rapidly inferring its physical properties, such as size, shape, and material composition, is essential. These characteristics offer vital clues about the object’s functionality, operational role, and potential threat, enabling a more informed assessment and response in an increasingly crowded space environment. To address this challenge, we have implemented a framework for simultaneous estimation of orbital state and physical target properties of unresolved and partially resolved resident space objects. The approach employs an augmented-state filtering architecture to estimate shape descriptors, material properties, and their uncertainties for SDA scenarios in the GEO and cislunar regime.
This work integrates a radiometric processing pipeline with GPU-accelerated path tracing to simulate realistic illumination and light transport reaching a sensor. This pipeline uses stochastic progressive photon mapping (SPPM) to efficiently accumulate photon energy on visible surface points, capturing contributions from both nearby areas on the target and global illumination effects from the Sun, Earth, and Moon. The photometric measurement output produces high-fidelity estimates of integrated photon flux, brightness, and signal-to-noise ratio (SNR) at the detector. These photometric outputs are processed in conjunction with angles-only optical measurements, allowing the filter to continuously refine both the orbital state and the augmented physical parameters in a sequential manner. This methodology is particularly effective in scenarios with sparse sensor tasking, where real-time estimation of unknown target geometry and material properties is critical.
The proposed framework holds significant relevance for safety and defense applications, as it enhances object tracking and identification capabilities in challenging observational conditions. By providing more accurate and timely situational awareness, this approach would enable decision-makers to act with greater confidence, supported by enhanced insight into object behavior, characteristics, and potential intent.
Date of Conference: September 16-19, 2025
Track: Satellite Characterization