Raymond Wright, BAE Systems; Geoffrey Lake, Icebox Engineering; Thomas Drouillard, BAE Systems; Andrew Wernersbach, BAE Systems; Kedar Naik, BAE Systems; Michael Dittman, BAE Systems; Matthew Tooth, BAE Systems
Keywords: Space Domain Awareness, Hyperspectral, Characterization, Space-Based Assets, Cislunar
Abstract:
The ability to rapidly identify Resident Space Objects (RSOs) from intrinsic signatures is critical for complete Space Domain Awareness (SDA) and achieving space superiority. Hyperspectral Imaging (HSI) from visible to long-wave infrared can detect signature differences between RSOs that are created by material composition, manufacturing, and geometric differences. These differences are exploited using Machine Learning (ML) algorithms to enable singleobservation identification of known objects and characterization of unknown objects. Space-based HSI systems provide access to critical wavelengths necessary for identification and characterization over ground-based HSI systems that are limited by atmospheric conditions, especially absorption features in frequencies where discriminating information lies. In this paper, the advantages of HSI are discussed, including its ability to provide a more comprehensive view of space objects compared to traditional single-band or multispectral methods. The paper details the benefits of space-based HSI, including a quantification of the wavelengths that are important for identification and cannot be detected from ground-based systems. The paper explores the design and trade space of a space-based hyperspectral sensor, included parameters such as range, aperture, and spectral resolving power, and presents a 51-channel, ultra-wideband HSI system that can detect RSOs as small as 30cm, up to 10,000km range, and is composed of high Technical Readiness Level (TRL) components. The paper then details the modeling and simulation used to demonstrate the effectiveness of the sensor with ML algorithms in two experiments: identification of individual satellites of the same model, and characterization of an unknown satellite’s material composition and mission. The experimental results show that the HSI sensor and ML algorithms can accurately identify individual satellites within the same geometric design and different material compositions with greater than 89% accuracy for measured cases from a single HSI observation. The characterization experiment on realistic simulated data shows that the material composition of an unknown satellite geometry can be identified with greater than 74% accuracy. This study shows that a new capability to identify satellites and characterize their missions in a single observation could be achieved, and lead to greater space domain awareness.
Date of Conference: September 16-19, 2025
Track: Space-Based Assets