Development and Application of Genetic Algorithms for Sandia's RATLER Robotic Vehicles

This report describes the development and application of genetic algorithms for the purpose of directing robotic vehicles to various signal sources. The use of such vehicles for surveillance and detection operations has become increasingly important in defense and humanitarian applications. The computationally parallel programming model as implemented on Sandia’s parallel compute cluster Siberia and used to develop the genetic algorithm is discussed in detail. The model generates a computer program that, when loaded into a robotic vehicle’s on-board computer, is designed to guide the robot to successfully accomplish its task. A significant finding is that a genetic algorithm derived for a simple, steady state signal source is robust enough to be applied to much more complex, time-varying signals. Also, algorithms for significantly noisy signals were found to be difficult to generate and should be the focus of future research. The methodology may be used for a genetic programming model to develop tracking behaviors for autonomous, micro-scale robotic vehicles.