Sensor fusion for robot navigation

Abstract

This paper demonstrates how a relatively simple vision analysis algorithm can be used in conjunction with sonar sensors to greatly improve the ability of the robot to perceive and accurately locate obstacles. So far, the system can only handle somewhat constrained environments: tall, rectangular, darker colored obstacles with well lighted, ligher colored walls forming the background. The intent was to demonstrate how a simple sensor fusion system can lead to great improvements in robot perception. The program is executed much more quickly than other more sophisticated 3D vision techniques. It is more reliable than if just one type of sensor was used. In addition, the robot, HERMIES-II, is capable of accurately maintaining its position and orientation in the navigation area. The navigation program is relatively inexpensive to implement, and is goal oriented. It allows relatively optimum paths to be found in a previously unexplored, potentially dynamic environment. 9 refs., 6 figs.