Intelligent Techniques for Star-Pattern Recognition

Abstract

This work presents the study of two different approaches for the attitude determination of space vehicles. The Neural Network approach is based on a simple Kohonen network, where the characteristics of a star distribution within the Field Of View (FOV) are matched against an on-board stored star map. The second approach utilizes a Genetic Algorithm (GA). The GA minimizes the discrepancy between the characteristics of the stars inside the FOV and a candidate FOV selected from the star map. Preliminary simulations indicate that both approaches work well and deliver good accuracy in determining the bore sight direction of the space vehicle with respect to the star map.