Therefore, most buildings do not have any valuable digital twin.Įxisting capturing technologies such as laser scanning and photogrammetry make it possible to collect efficiently point clouds that contain geometric information about the as-is state in the built environment. The other reason is that even though some new buildings have a digital design model, this model was not updated when the asset was modified throughout its life cycle. There were no authoring tools for constructing a three-dimensional (3D) digital model or the concept of digital twin when they were built. The first one is that many buildings were constructed decades ago. There are mainly two reasons for this situation. However, despite a digital twin being valuable for building management and maintenance, only a few existing buildings have a reliable basic digital twin that only contains structural elements and reflects their current as-is state. A geometric digital twin means a digital representation that contains the current geometric information on physical assets it refers to. This paper focuses on creating geometric digital twins of buildings. A digital twin is a regularly updated digital replica of a physical built asset in a built environment ( Brilakis et al., 2019). In architecture, engineering, construction (AEC) and facility management sectors, digital twins have been entering the conversation, as they can continuously offer substantial value to all associated stakeholders in various aspects. Room space of room i in the manually created building information modelling model Ratio value for checking stopping condition in the top direction Ratio value for checking stopping conditions in other directions Ratio value for checking stopping conditions in the bottom direction Ratio value for checking geometric stopping conditions in a direction Room space of the extracted method for room i Number of non-void voxels in the top direction Number of non-void voxels in the other directions Number of non-void voxels in the bottom direction Number of void voxels in the same direction ![]() Number of voxels predicted as a window in the other four directions Number of voxels predicted as a wall in the other four directions Number of voxels predicted as the floor in the bottom direction Number of voxels predicted as a door in other directions Number of voxels predicted as the ceiling in the top direction Number of non-void voxels in one direction By taking useful semantic information into consideration, the proposed approach performs better in creating geometric digital twins of buildings. This study used existing state-of-the-art deep learning architecture for the segmentation task in the proposed approach. Compared with previous studies that mainly use geometric information only, the approach also focuses on how to select useful information predicted by deep learning. The method can work in ( a) rooms with complex structures like U-shape and L-shape, ( b) rooms with different ceiling heights and ( c) rooms under a high occlusion level. Then, based on the detected room spaces, structural elements, as well as doors and windows, are extracted. Unlike most previous research that starts with detecting planes in the point cloud and considers only geometric information, the proposed ‘void-growing’ approach is a full-automatic approach that starts with detecting void space inside rooms, considering geometric information, as well as semantic information predicted from deep learning. The challenge that this paper addresses is how to generate geometric digital twins of the indoor environment of buildings automatically.
0 Comments
Leave a Reply. |