Our paper titled “Navigation in AR based on Digital Replicas” has been accepted in the Computer Graphics International (CGI) 2018 conference. The same is authored by Ojaswa Sharma, Jalaj Pandey, Hammad Akhtar, and Gaurav Rathee. The work will be published as a full paper in the Visual Computer Journal.
In this paper, we address the two main topics of 3D navigation and space identification within the context of mixed reality. Creating navigable digital replicas from real-life buildings is a cumbersome task. We present a mostly automated pipeline to process 3D geometry created from architectural blueprints. We discuss a coherent procedural approach to build the topological information required for navigation, and a semi-automatic generation of hierarchical tags for identification of spaces. The geometric and topologic information along with tags are stored in a spatial database. We address challenges in automating the entire process such that manual effort is reduced to minimal. Our approach to asset creation enables navigation and identification in both indoor and outdoor spaces. Such a digital infrastructure is central to any VR and AR system that utilises these assets for further computations.