Fast and robust kernel generators for star trackers
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
2017 IAA Star Identification (Star-ID) is a complex problem, mainly because some of the observations are not generated by actual stars, but by reflecting debris, other satellites, visible planets, or by electronic noise. For this reason, the capability to discriminate stars from non-stars is an important aspect of Star-ID robustness. Usually, the Star-ID task is performed by first attempting identification on a small group of observed stars (a kernel) and, in case of failure, replacing that kernel with another, until a kernel made only of actual stars is found. This work performs a detailed analysis of kernel generator algorithms, suitable for on-board implementation in terms of speed and robustness, for kernels of three (triad) and four (quad) stars. Three new kernel generator algorithms and, in addition to the existing expected time to discovery, three new metrics for robustness evaluation are proposed. The proposed algorithms are fast, robust to find good kernels, and do not require pre-stored data.