Creating instructional content for augmented reality based on controlled natural language concepts

In this paper we present an approach to support the content generation for Augmented Reality (AR)-based technical manuals by modeling and interpreting technical instructions in a semantical way. We propose the separation of the meaning of a technical instruction from its actual representation and accomplish this by designing a semantic instruction model on the base of a controlled natural language. The controlled language serves here as a high-level specification model for technical activities. Its controlled lexicon and grammar eliminate the ambiguity of meaning and reduce the complexity of language constructs known from full natural languages. This enables automated processing of the instructions. Complemented with references to the geometry of assemblies, parts, tools and other geometric features, the resulting formalized instructions can be easily transformed to both textual and graphical representations. In the demonstration of our results we will focus on the generation of scenes for AR-based maintenance support.