Autonomous robot navigation using human spatial concepts

With the aim of automatically reasoning with spatial aspects in a cognitive way, several qualitative models have been developed in recent years. However, there is no model to reason with several spatial aspects in a uniform way. Moreover, most of these models simplify spatial objects to points. In this paper we present the use of constraint logic programming instantiated to finite domains extended with constraint handling rules as a tool for successfully integrating the qualitative concepts of orientation, distance, and cardinal directions, using points as well as extended objects as a primitive of reasoning. The resulting model has been applied to build a demonstrator: a qualitative navigation simulator on the structured environment of our city. (C) 2000 John Wiley & Sons, Inc.