SSATT: A TOOL FOR AUTOMATED EVALUATION OF STAR SENSOR DESIGN, PERFORMANCE AND ON-BOARD ALGORITHMS

摘要

In recent years, many star sensor developments have started up. This has been the incentive for development of general W95 tool to assist design and verification of star sensor hardware and algorithms: the SSATT or Star Sensor & Algorithm Test Tool. Performance is measured in obtainable attitude accuracy, reliability, database size and recognition time. To allow for a quantification of performance and competitive comparison of results, new algorithms have been developed for automated on-board database construction, pointer based database search, star triangle recognition and validation techniques. The newly developed algorithms have been tested against leading existing algorithms (Liebe, Quine, Van Bezooijen) and show excellent performance. Recent work has been the development of algorithms for detection and centroiding and their optimization for a specific camera or lens. This allows for a direct connection between real hardware and the SSATT algorithms. The tool has been very successfully applied to synthetic images provided by TNO-TPD as well as real imagery taken from the web. At the moment, the tool is limited to the Lost In Space problem, however current work involves extension to attitude propagation, rate determination, tracking, and use of a priori-knowledge. The algorithms included can be easily transferred to flight software, while on the other hand existing recognition algorithms can be readily coupled to the Monte Carlo tool for performance comparison. A demo version is available via the web-site This paper describes the functionality of the SSATT, discusses the algorithms that are included and shares some investigation results.