Spatially Constrained Grammars for Mobile Intention Recognition

  • Authors:
  • Peter Kiefer

  • Affiliations:
  • Laboratory for Semantic Information Technologies, Otto-Friedrich-University Bamberg, Bamberg, Germany 96045

  • Venue:
  • Proceedings of the international conference on Spatial Cognition VI: Learning, Reasoning, and Talking about Space
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Mobile intention recognition is the problem of inferring a mobile agent's intentions from her spatio-temporal behavior. The intentions an agent can have in a specific situation depend on the spatial context, and on the spatially contextualized behavior history. We introduce two spatially constrained grammars that allow for modeling of complex constraints between space and intentions, one based on Context-Free, one based on Tree-Adjoining Grammars. We show which of these formalisms is suited best for frequently occurring intentional patterns. We argue that our grammars are cognitively comprehensible, while at the same time helping to prune the search space for intention recognition.