Abstract

An uncertainty management subsystem for a mobile robot is described within the context of the autonomous robot architecture (AuRA). This system incorporates a spatial uncertainty map that represents both the positional and orientational uncertainty in a mobile vehicle relative to a global map. The spatial uncertainty map consists of a convex polygonal region and a compass wedge which are used to relate the robot's current position relative to the world model.

Techniques for uncertainty growth during motion and uncertainty reduction by landmark recognition are presented. The spatial uncertainty increases as the robot moves through the world in a manner dependent on the underlying terrain. Three different landmark classes are described, each having a different impact upon the reduction of uncertainty. Expectations generated by the spatial uncertainty map are used to constrain the perceptual processing required for landmark recognition. Experimental results using our mobile robot demonstrate the viability of this method.

Author Keywords: spatial uncertainty; mobile robot; expectation-based perception; robot navigation; landmark discovery and tracking

Article Outline

• References

Corresponding author. Address correspondence to Ronald C. Arkin, School of Information and Computer Science, Georgia Institute of Technology, Atlanta, GA 30332.