Optimization of a Sun-Sensor Illumination Pattern using Genetic Algorithms

Changes to the illumination pattern in a digital sun-sensor can dramatically improve the resolution of the device. In this study we use genetic algorithms (GAs) as a heuristic to optimize the illumination pattern for a single-axis digital sun-sensor. The main objective is to determine an illumination pattern that resolves the sun-angle to better than one pixel. A linear-phase super resolution technique is proposed, to evaluate the effective resolution of the illumination pattern determined using GA. Our finding show that patterns with multiple, narrow peaks provide sub-pixel accuracy in resolving the sun-angle. Performance of the proposed GA estimator displayed the evolution of high-fitness solutions. We contend that multiple peak patterns can greatly improve the performance of the sun-sensor when coupled with parametric methods of displacement estimation. The optimal illumination pattern can be implemented by fabricating a replacement aperture mask for the sensor - a change that can be made at minimal cost