Measurement-based Birth Model for a Space Object Cardinalized Probability Hypothesis Density Filter
Brandon A. Jones, Steve Gehly, Penina Axelrad · AIAA/AAS Astrodynamics Specialist Conference · 2014
The cardinalized probability hypothesis density (CPHD) multi-target filter allows for the state estimation of known and previously unknown objects in challenging observation environments. To identify new objects, the CPHD requires a birth model to define the density of potential new targets in the single-target space. This paper presents an observation-based birth model using the constrained admissible region that identifies previously unknown space objects and instantiates a Gaussian mixture representation of the new target density. The Gaussian mixture formulation of the CPHD then refines the state estimate of the new target while simultaneously tracking known targets. Two variations of a simulation test demonstrate the efficacy of the method for the tracking of objects in the near-geosynchronous region using groundand space-based optical sensors. The first test case shows the filter’s performance for dense clutter and a priori knowledge of the new target’s presence. The second variation demonstrates the filter’s ability to track known and unknown objects with no a priori knowledge of the new target and a reduced clutter density. This latter case demonstrates that the CPHD with the new birth model correctly identifies potential new targets, discards false detections, and maintains custody of the previously known objects in the multi-target state.