Joseph Diamond, Peraton; Michael Goodman, Peraton; Peter Kent, Peraton; Stephen Sawyer, Peraton; Sasidhar Devabhaktuni, Pennsylvania State University; Roshan Eapen, Pennsylvania State University; Jonathon Hope, Pennsylvania State University; Puneet Singla, Pennsylvania State University
Keywords: Peraton, Penn State, PSU, Lockheed, FireOPAL, Cislunar, xGEO, IOD, Classification
Abstract:
A sophisticated understanding of motion of a spacecraft in multi-body environments such as the cislunar space is essential to forecast and track objects for persistent surveillance, mission safety, and collision avoidance, and adversarial behavior. The primary challenge that limits the transferability of tools and techniques from the LEO/GEO to xGEO regions is non-Keplerian dynamics, data sparsity from limited coverage and availability of sensors. Furthermore, objects in the cislunar region suffer from few astrometric observations with large measurement cadences, and in many cases, the amount of information is too limited to compute a full orbit according to the least squares principle. This paper will investigate if such measurements contain significant orbit information allowing rapid classification of trajectory: LEO-to-GEO or xGEO. For example, electro-optical observations at large distances typically require very long exposures or stacking of multiple sequential images. In such situations, one can reasonably estimate the angular rates of the object on that trajectory. Of particular significance to this work is the FireOPAL (Fireball OPtical ALert) observatories located in Australia, operated by Lockheed Martin Australia. The field of view for this observatory system is 20×12.5 deg, and are designed to function as a coherent network, to monitor a large number of artificial satellites simultaneously, and provide wide area surveillance and precision tracking in near real time. Satellite measurements from the FireOPAL system are very suitable for extracting angle rates through either post processing, or multiple measurements. Furthermore, if we can constrain the space of range-range-rate to distinguish between satellites in orbits around Earth and xGEO object, an admissible region can be constructed to classify different type of trajectories. We will be utilizing the concept of admissible regions earlier defined for asteroid identification. This work intends to simulate the sensor model for the FireOPAL system to be used in the rapid classifier and further validate the work with real data from FireOPAL. Thus, the goal of this research is to develop a rapid classifier for near-Earth and xGEO objects based on electro-optical measurements and image processing.
Methodology: Given an observation, O, the following conditions are considered which have obvious physical interpretation: (a) D1: O is not a planetary body, asteroid, or comet; (b) D2: The trajectory of O is influenced by both Earth and Lunar gravity; (c) D3: The trajectory of O goes beyond GEO. One can impose further conditions that distinguish near-Earth and cislunar trajectories. The admissible set then contains the union of D1, D2, and D3. Insighted from two-body trajectory representations and a restricted three-body dynamical model will be utilized to construct this admissible region. The admissible region being in the range-range-rate space coupled with the information on angles and angle-rates provides sufficient information to reconstruct potential initial conditions that explain the trajectory. If the angle and angle-rate can be extracted from a single observation, the characterization can be immediate. Else, one must wait for additional observations. It will also be explored how the admissible region can be shrunk further by incorporating information of multiple measurements. By incorporating statistical filtering techniques and optimization methods, the process can be further refined to reduce uncertainty and improve prediction capabilities. A combination of synthetic data as well as the real data collected from FireOPAL will be used to validate the developed methodology. Ultimately, these developments will contribute to a more robust framework for rapid-classification and subsequent initial orbit determination (IOD) methods for objects in the near-Earth and cislunar environment.
Expected Outcomes: The following contributions are envisioned: (a) Developing a quick classifier to identify xGEO objects from the observations near the Earth, (b) developing a framework to generate synthetic optical data capturing true dynamics, camera and telescope properties, and to evaluate a probability of detection (which can be used as an additional parameter in the classification), (c) Utilization of FireOPAL data- a distributed network of standalone, fixed mount optical sensors that is designed to monitor a large region of the sky- for validation of developed technologies, (d) a robust pipeline for cislunar space domain awareness.
SDA Applicability: The proposed framework can be used as an early-warning system of potential cislunar objects and prompt the use of additional sensors to persistently track them. One can maintain a compact catalog, in terms of admissible region variables, of identified cislunar objects. It has numerous benefits to IOD algorithms providing a good initial guess for algorithms such as nonlinear least-squares to initiate.
Date of Conference: September 16-19, 2025
Track: Cislunar SDA