Dynamic composition of interactive AR scenes with the CARL language

The paper presents a method of dynamic composition of interactive augmented reality (AR) scenes based on distributed data sources and depending on the context. For this purpose, a novel declarative language for modelling ubiquitous, contextual AR environments is employed. The language, called CARL - Contextual Augmented Reality Language, enables modularisation of the structure of AR scenes and the presented AR content. CARL separates specification of three categories of entities constituting an AR environment - trackable markers, content objects and interfaces. Particular entities may be selectively added to an AR scene, which makes the language flexible and particularly well suited for building dynamic AR systems.