Lost-in-space: A star pattern recognition and attitude estimation approach for the case of no prior attitude information
Conference Paper
Overview
Identity
Additional Document Info
View All
Overview
abstract
A novel attitude determination approach is presented and results from night sky validation tests are discussed The central ideas are extensions of the K-Vector method recently introduced by Mortari. It involves a construction of a judicious "star pair catalog" prior to launch, wherein all cataloged stars are considered as star pairs and are ordered {k = 1, 2, .... NO(106)}, over the whole sky, sorted in the order of increasing inter-star angle. From this, we have a "searchless" means to access the candidate set of stars for each measured star pair, but this K-Vector method may still give 10s and sometimes 100s of candidate stars for each measured pair. We introduce in this paper a method to identify the measured stars in the subset of candidate stars accessed using the K-vector. The new method is based upon a logical process which pivots about two stars (the first identified pair) to more efficiently identify or ignore the remaining measured stars. Using this Pivot Method, we show that star identification can be reliably and efficiently implemented on-orbit, even for the Lost-In-Space case, and thereby this paper introduces a globally valid process, consistent with real-time, on-board computational constraints, that solves the most fundamental problem associated with star pattern identification. Results from night sky experiments are discussed which support the validity of the analysis and practicality of this approach. Also discussed are plans for on-orbit experiments. The StarNav experiment is planned for Shuttle Mission STS 107, January 2000; this will represent the first ever on-orbit demonstration of a star sensor implementing a Lost-In-Space star identification and attitude determination process.