IRONCASTLE Interpretation and Recognition of Objects using Novel Coupled Attitude and Shape Techniques for Lightcurve Evaluation
Abstract
There exists an important need to characterize space objects beyond their orbits and centers of mass. One readily available way of collecting relevant information is via the use of light curves: a time series of non resolved brightness measurements of the object. This information can support the determination of the attitude and albedo shape through inversion. Current methods fail in the absence of exact partial information on attitude or albedo shape. However, even this partial information is usually absent. Simultaneously, by its very nature, the inversion is severely under determined. Hence, without constraints based on some prior knowledge, it is not possible to converge to a single solution even for geometrically simple objects. In this new approach, a methodology is developed, and validated on synthetic light curves, where the "ground truth" is known, and on real measured light curves. The methodology is centred on a Bayesian Multi Hypothesis approach, which will map out possible solutions to the under determined inversion as sets of possible attitude albedo shape hypothesis and assign weights to support the validity of each hypothesis. Weights are computed based on constraints imposed by the underlying physics and through the addition of limited available supplementary information including orbital information, additional light curves, and eventual spectra. The process of determining the initial weights will be performed with the aid of a Hopkins neural network, developed exclusively for this data fusion task. Ultimately, this will lead to a limited number of highly weighted hypotheses, which will allow for informed decision making about the most likely attitude albedo shape solution. Finally, a feed forward neural network will aid the interpretation of this optimal solution, in terms of the operational status and behaviour of the space object.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 14, 2022
- Source ID
- FA95501910407
Entities
People
- Carolin Frueh
Organizations
- Air Force Office of Scientific Research
- Purdue University
- United States Air Force