Marcus J. Holzinger (School of Aerospace Engineering, Georgia Institute of Technology and Aerospace Engineering Department, Texas A&M University), Kyle T. Alfriend (School of Aerospace Engineering, Georgia Institute of Technology), Charles J. Wetterer (Pacific Defense Solutions, LLC), K. Kim Luu (Air Force Research Laboratory, USAF), Chris Sabol (Aerospace Engineering Department, Texas A&M University), Kris Hamada (School of Aerospace Engineering, Georgia Institute of Technology), Andrew Harms (Department of Electrical Engineering, Princeton University)
Keywords: Non-Resolved Object Characterization
Abstract:
The increasing number of manufactured on-orbit objects as well as improving sensor capabilities indicate that the number of trackable objects will likely exceed 100,000 within the next several years. Characterizing the large population of non-spatially resolved active spacecraft, retired spacecraft, rocket bodies, debris, and High Area to Mass Ratio (HAMR) objects necessarily involves both attitude and shape estimation. While spatially unresolved space objects cannot be directly imaged, attitude and shape may be inferred by carefully examining their lightcurves. Lightcurves are temporally-resolved sequences of photometric intensity measurements over one or more bandwidths. Because the observable reflected light from an unresolved space object is a strong function of both its shape and attitude, estimating these parameters using lightcurves can provide an avenue to determine both space object attitude and shape. This problem is traditionally called `lightcurve inversion.’ While lightcurves have been used for 25 years to characterize spin states and shapes of asteroids, estimating the attitude states and shapes of manufactured space objects involves a new set of challenges. New challenges addressed in this paper are 1) An active (agile) space object is often directly controlling its attitude, meaning that torques acting on the space object are not necessarily zero (non-homogeneous motion) and mass properties may not be known, 2) Shape models must often be estimated, and as such contain errors that need to be accounted for in the measurement function, 3) Dynamics and measurement functions are excessively nonlinear, and manufactured space objects may be quite symmetric about at least one axis of rotation/reflection. This can lead to multiple possible attitude estimate solutions and suggests the use of non-Gaussian estimation approaches. Agile space objects (those that can actively maneuver) pose new problems to lightcurve inversion efforts to estimate attitude. Because the torques acting on a maneuvering space object are in general non-zero, it is very difficult to decouple the normalized mass properties from the efforts of the space object actuators. This paper decomposes instantaneous angular rates into mean and temporally random components and leverages the resulting behavior to track slewing space objects. Models of the space object shape are a necessary input for attitude inversion efforts. However, in operational settings the true shape model is never known exactly (even when a CAD model is available), and is often approximated using facet models with fitted surface parameters. Using estimated shape models without accounting for the discrepancy with truth can often result in filter divergence, particularly under glint conditions. This paper addresses how differences between the true and estimated shape models may be rigorously accounted for and deleterious effects mitigated using shape model bias dynamics. Bayesian estimation methods, specifically Particle Filters, are used to account for severely non-Gaussian distributions resulting from both nonlinear dynamics and measurement functions.
Date of Conference: September 11-14, 2012
Track: Non-Resolved Object Characterization