“System on a Chip – Realization of Pattern Recognition Genetic Algorithm using a Microcontroller”

Jason Marenda, Marco P. Schoen, Larry Stout, and Brian G. Williams
Idaho State University


Uninhibited spacecrafts are used in many different technology sectors, including the communications and military sectors.  The attitude determination for these spacecrafts generally makes use of a star tracker.  The star tracker used must be able to successfully perform the pattern recognition algorithm quickly and accurately.  This has been accomplished in a previous work by the authors using a multi-objective genetic algorithm (MOGA).  The algorithm utilizes an on board stored star map; because star field pattern recognition is more accurate than previous single or bright star trackers.  The algorithm was first optimized and developed for use with a personal computer resulting in attitude determination accuracy above 99%.  Building on the success of the results using a personal computer, the resulting optimal algorithm was realized using a microcontroller.  The microcontroller accepts an input from a camera in the form of a three dimensional array which represents the star field view.  The algorithm then compares the input star field to the stored star field map to determine the spacecraft’s attitude, and the spacecraft’s attitude is output not only via the microcontroller’s output ports, but also via a serial port.  The realization of the algorithm using a microprocessor is the next step towards achieving a complete attitude determination system on a chip.  The current performance characteristics, including timing and accuracy, are discussed.  The paper concludes with the discussion of the future of the project.

2006 MAPLD International Conference Home Page