Leonardo Cament, Universidad de Chile; Martin Adams, Universidad de Chile; Elías Obreque, Northumbria University; Pablo Barrios, Universidad de Chile
Keywords: SSA, multi-object tracking, Poisson Labelled multi-Bernoulli Filter, Dynamic Orbit Model
Abstract:
Tracking and cataloging orbiting Space Objects (SOs) represent a cornerstone of modern Space Situational Awareness (SSA). The inherently stochastic nature of orbital dynamics, influenced by multiple external perturbations, necessitates the use of probabilistic filtering methods for reliable state estimation. The predominant focus of prior research has been on Gaussian Mixture (GM)-based techniques or on particle filters, the latter often associated with high computational cost. Nevertheless, GM formulations are insufficient to capture the non-Euclidean geometry intrinsic to orbital motion. In this work, the Probabilistic Admissible Region (PAR) method is applied for initial orbit determination. Furthermore, a manifold-based framework is introduced as an alternative to conventional GM representations. By modeling the SO state as an element of a Riemannian manifold M, the proposed approach enhances predictive capability while retaining computational tractability. ACubature Kalman Filter (CKF)-based predictor is then formulated on M, and its performance is evaluated using orbital measurements from four CubeSats operated by the Universidad de Chile. Preliminary results demonstrate that the proposed strategy achieves performance comparable to linear Gaussian prediction in short-arc scenarios and, in certain cases, superior accuracy over longer observation intervals.
Date of Conference: September 16-19, 2025
Track: Astrodynamics