A genetic algorithm for robot navigation

Abstract

The sensory capabilities of the human brain makes it possible for even blind people to maneuver through unfamiliar terrain with the help of guiding devices. However, the abilities of modern robots are not so sophisticated. Given a terrain with obstacles, determining an optimally short robot navigation path poses a major search problem. In this paper, a genetic search method is developed to solve the robot navigation problem. The search node or path is represented by a string of integers, each integer representing a cell in the terrain. Simulation results comparing the proposed genetic algorithm and a sequential navigation algorithm are presented.