The attitude control and navigation systems of future advanced spacecraft will be characterized by a high degree of autonomy, very high accuracy, efficient commandability, and fast fault recovery. These characteristics are incompatible with the constraints of conventional star sensors which mandate a-priori definition of all onboard attitude fixes and work only if attitude uncertainties remain small. With the availability of accurate, anti-blooming capable CCDs, fast microprocessors, high density memory chips, and star pattern recognition algorithms, it is now feasible to fabricate miniature Autonomous Star Trackers (ASTs) capable of (1) determining their attitude rapidly and reliably while having no a-priori attitude knowledge, (2) autonomous attitude updating, and (3) providing their attitude at rates up to typically 40 Hz. In addition to providing the functionality needed for future missions, ASTs can also be exploited to improve the reliability, mass, power, and cost of spacecraft and reduce the cost of operating them.
This paper describes star identification schemes used in the past, it discusses a number of star pattern recognition algorithms, and provides the main characteristics of current CCD star trackers. A number of specific functions enabled or enhanced by an AST are described including fast attitude acquisition, rapid fault recovery, attitude safing, gyroless/cheap-gyro attitude control, autonomous target acquisition by astronomy telescopes, autonomous optical navigation, and precision pointing to terrestrial targets. The AST being developed at Lockheed uses a fast, memory-efficient, and highly robust star pattern recognition algorithm based on matching groups of stars. The algorithm, which is also applicable to star scanners, is described along with a realistic simulation program for testing its performance. It is shown that an AST with an 11.3 degree FOV diameter, a database of 4100 guide stars, a 25 MHz MC68030 class microprocessor, and 800 Kbytes of memory will be capable of determining its attitude in 0.45 seconds with a success rate greater than 99.98% when using an optimal guide star selection method. Compute time and memory are found to be inversely proportional to the FOV area. The paper also reports on AST development by other organizations in a number of countries.