Seamless Augmented Reality Registration Supporting Facility Management Operations in Unprepared Environments

Despite its great potential, Augmented Reality (AR) still struggles to be widely used in real processes in the construction industry. This is mainly due to limitations associated with current AR registration methodologies, including the lack of continuity between different scenarios, drifts over distances, and the need to prepare the scene in advance with alignment infrastructures and to resort to manual registration procedures. In addition, users may not be skilled enough or allowed to prepare the considered environment, since this is a task that requires a significant amount of measurement and calibration. To promote the application of AR in Facility Management (FM), which typically involves both complex indoor and outdoor environments, an automatic localization method is needed, thus providing accurate AR registration in mixed and unprepared environments. In order to fill this gap, a system implementing a markerless approach for seamless indoor-outdoor AR registration has been developed by integrating multiple AR registration engines with a cloud platform. These engines, primarily relying on Global Navigation Satellite Systems Real-Time Kinematic (GNSS-RTK) and Computer Vision (CV) technologies, are managed by an additional system component that automatically assigns priority depending on the current scenario. The system proposed in this study is tested on site on an FM use case related to a university campus for qualitative assessment. Furthermore, a quantitative assessment of the system’s accuracy is also performed. In both evaluations, the system shows very promising results in terms of (i) applicability for FM operations, and (ii) accuracy. Specifically, BIM holograms are automatically superimposed to their real counterparts seamlessly, in mixed indoor-outdoor scenarios, without manual procedures, demonstrating applicability of the proposed approach in unprepared environments. In terms of accuracy, the AR overlaying discrepancies in outdoor and indoor scenarios (D_(L-SQ)) result equals 0.090 m and 0.082 m with a maximum among all the analyzed scenes (D_M) of 0.075 m and 0.071 m, respectively. Given these results, the work presented in this paper provides a substantial contribution to the dissemination of AR technologies in the FM field and to the support of FM activities in terms of better coordination, visualization, communication and on-field access to high-quality information. This paper extends the work presented at the 23rd International Conference on Construction Applications of Virtual Reality (CONVR 2023).