Stochastic Star Identification.pdf

., Stochastic Star Identification Suraphol Udomkesmalee,’ James W. Alcxancler,f’ and Aurelio F. Tolivar$ Jet Propulsion Laboratory, California Institute of Technology Pasadena, California 91109 Al)stract An approach to star identification based on comparing observed pattern statistics with the precomputed star cataloged statistics is suggested. ~’he identification criterion is based on evaluating a posteriori probabilities of designated star sequences obtained from observing different star fields. Numerical results based on a specific algorithm are presented. A number of references for other approaches are cited. Introduction Operation and performance of spacecraft control systems will be significantly improved with the ability to autonomously and precisely calctilate spacecraft attitude by means of star sensors. With recent advancements in charge-coupled device (CCL)) star trackersl~ and NASA’s emphasis on developing cost-effective spacecraft, a star tracker with full-sky recognition based on multiple star observations will have a significant impact in terms of cost, limitations, and operational complexity of spacecraft and space-based instruments. The ability to identify guide stars autonomously will provide more efficient, robust, and independent modes of attitude determination and fault or loss- of-attitude recovery. Once the star field is correctly identified, spacecraft attitude can be determined by generating the direction cosine matrix from one pair of identified stars or solving for the attitude with the least- squares method using all the identified stars in the field of view (FOV),5 or various other methods. Having established the initial attitude, subsequent star pattern recognition and attitude determination processing will be greatly simplified (tracking known stars becomes an issue here). Although many algorithms have beeen proposecl for star pattern identification, significant limitations in terms of parametric sensitivities, flexibility, and i