Affiliation:
1. Università Politecnica delle Marche, IT
Abstract
Augmented reality (AR) still struggles to be widely used in real processes in the construction industry despite its great potential. This is partly due to the difficulties that exist in aligning holograms and maintaining their stability, especially for outdoor applications. In addition, being indoor-outdoor interactions crucial for built environment management, it would be important that AR apps can work seamlessly. Alignment in indoor environments cannot make use of methods such as GNSS, nor can all environments be assumed to have been previously initialized with AR tools. Thus, marker-less AR registration is crucial for indoor applications. This paper presents an approach for marker-less AR registration seamlessly in both outdoor and indoor environments. Real-time kinematic positioning (RTK) and Inertial Measurement Units (IMU) technologies have been chosen for outdoor registration, while image comparison based on convolutional neural networks (CNN) for indoor registration. In this research, the application of these two technologies and their integration have been studied and tested on site on a real Facility Management use case related to a university campus. The proposed approach has shown very promising results in displaying BIM elements of the electrical system seamlessly superimposed through AR to their physical counterparts in mixed indoor-outdoor environments