Development and Application of Genetic Algorithms for Sandia's RATLER RoboticVehicles

摘要

The emerging technical approach to deal with a challenging, possibly hostile,environment is likely to involve a large number of small, but fairly intelligent, robots. It is envisioned that these can covertly infiltrate a designated area, enter buildings, gather appropriate information, and communicate with and learn from each other. They would also communicate with a smaller number of on-the-scene soldiers backed up by powerful off-line computers that can carry out large-scale information collection, analyses, and simulations. This report documents the effort to generate and apply a robust genetic algorithm to act as a vehicle controlling program for robotic behavior. In a typical scenario, robots are initially distributed randomly in a field and given the goal of locating the emitting source, be it sound or smell.