DFKI-LT - Interpreting Motion Expressions in Route Instructions Using Two Projection-Based Spatial Models
Interpreting Motion Expressions in Route Instructions Using Two Projection-Based Spatial Models
KI 2008: Advances in Artificial Intelligence volume 5243,
Lecture Notes in Artificial Intelligence,
This paper explores the applicability of two formal models of spatial relations, Double Cross and RfDL 3 − 12, to interpret some typical expressions that people use for describing a route. The relations in these two models allow the qualitative representation of the location and spatial extent of a landmark as seen from a route segment. We explore the correspondence between the relations in these two models and the motion expressions that refer to a point-like and a region-like landmark, respectively, which consist of the same set of direction-related expressions and specific sets of topology-related expressions. Through this exploration, we identify intrinsic ambiguities in the direction-related motion expressions that refer to a region-like landmark. Finally, we propose the generalization of our approach by using a spatial ontology, which potentially enables the mobile robots to interpret a large variety of expressions in human route instructions.