As Augmented Reality systems and applications expand in scale and scope, so do the complexities of managing interactions and information they provide. Managing these systems requires a structured, yet flexible model. Based on the information-processing pipeline, a research project, funded by the Austrian Science Fund FWF, has designed a 4-tier structure for managing complex Augmented Reality models. (Schmalstieg, et al., 2007)
The article depicts 4 layers for managing complex AR models “acquisition, storage, delivery, and use of the data” (Schmalstieg, et al., 2007).
Tier 0: Acquisition
This refers to the acquisition of data to be used in the AR system/application. The options for this could range from legacy systems to information collected as the users interact with the real world, as well as, information generated/authored by the system itself.
Tier 1: Storage
This refers to the storage of acquired data from Tier 1. The central issue with data storage is designing a structure that is “sufficiently flexible” to service a wide variety of applications (Schmalstieg, et al., 2007). For this reason, a highly flexible, yet structured data model is required.
For this the team developed an “XML dialect called Building Augmentation Markup Language (BAUML)” to provide an flexible extensible structure. (Schmalstieg, et al., 2007).
Tier 2: Delivery
Augmented Reality, very frequently, uses content that is generated on-the-fly (at run-time) as it is sensitive to the “context” of the user. Being able to model and remodel this data to fit a user’s context reduces the need for data duplication and optimizes bandwidth utility.
Tier 3: Use
Defining the way in which users interact with the application is an important factor in AR. As, truly ubiquitous, AR systems are required to allow user interaction in nearly countless situations, it is not practical to set a fixed set of umbrella interactions (that affect all real-world objects). Interactions with real world objects, that are being “augmented”, needs to be as unique as the objects are to each other. It is fair to say, collecting a fixed set of interactions, for each and every real-world-object, would be near impossible and highly-impractical. For this reason, Interactions, like the data, will have to be modeled on-the-fly.
One approach would be to create modular interactions that are aggregated, based its relevance to the object and data, at run-time
This model emphasizes the use of structured, re-usable, modularized information and instructions that are composable at run-time. This is to ensure enough flexibility to take into account varying applications and interactions. It also serves to minimize the storage, processing and transmission overheads in the system. This structure will serve to provide scalability as AR systems move towards becoming truly ubiquitous.
References
Schmalstieg, D., Schall, G., Wagner, D., Barakonyi, I., Reitmayr, G., Newman, J., et al. (2007, JUL/AUG). Managing Complex Augmented Reality Models. Computer Graphics and Applications, IEEE , 27(4), 48-57. Available Here
Augmented Reality Today 