Point Cloud Information Modeling (PCIM): An Innovative Framework for As-Is Information Modeling of Construction Sites

Point clouds constitute a high fidelity representation of the 3D geometry of physical objects; thus, point cloud technology is often used to create an as-built building information modeling (BIM) of a building construction site. However, the conventional scan-to-BIM pipeline still requires significant time and manual efforts to complete. To address this problem, this study proposes a novel point cloud information management format, called point cloud information modeling (PCIM), where semantic information of building elements is captured from the laser scanner and camera data and directly encoded in a combined point cloud data structure. PCIM is a concise representation of the as-is condition of building elements on a construction site, including building object information such as type of building element, material, and geometry. PCIM has an object-oriented hierarchical structure, and the definition of building objects are derived from the industry foundation classes (IFC). The process of creating PCIM can be automated since the underlying semantic structures such as class and material information can be directly extracted from raw sensor data with machine learning technology. Under the machine learning framework, an automated classifier can be trained to recognize relevant entities from point cloud data based on a pre-built library of 3D building computer-aided design (CAD) objects. To validate the overall PCIM framework, this research conducted a case study at an actual building under construction. The test results demonstrate that PCIM can be an effective tool for as-is information modeling of structures and facilities during construction.