Constructing A Complete And Accurate As-Built BIM Based On An As-Designed BIM And Progressive Laser Scans
In the current practice, changes that occurred in the construction or renovation phase are not frequently captured and updated to as-built documentations. As a semantically rich digital representation, a building information model (BIM) can be used as an information repository to store and deliver as-built information. Since a building might not be constructed exactly as its design specifies, discrepancies might exist between BIMs created based on design information and actual building conditions. It is important to keep information stored in a BIM accurate and complete so that it servers as a reliable data source throughout the service life of a facility. Point cloud data captured by laser scans is capable of providing accurate 3D depiction of building conditions. Hence, point cloud data can be used as the reference to update the information contained in an as-designed BIM. There are two main challenges associated with the update process. First, in order to recognize the differences between a point cloud and an as-designed BIM, segments captured by the point cloud need to be matched to building components modeled in the BIM. However, a building component might have different shapes, dimensions, and locations in a point cloud as compared to how it is modeled in an as-designed BIM. The discrepancies between a point cloud and an as-designed BIM increase the complexity of matching the two data sets together. Second, occlusions and construction progress prevent a laser scan performed at a single point in time from capturing a complete set of geometric information associated with a building. While it is possible to retrieve more geometric information by combining point clouds captured at different points in time, the large file size of a combined point cloud make it difficult to process and store. Hence, there is a need for an approach to achieve the balance between the richness of the information provided by a combined point cloud and the file size of it. In order to address the above two challenges, this research aims to develop a framework that supports the update process by matching information from a point cloud to a BIM and leveraging point clouds captured at different times to provide a more complete and accurate reference. To develop such a framework, this research developed the following research objectives: (1) Identifying a general set of features that can be applied to recognize correspondences between a point cloud and a BIM. (2) Developing matching approaches that match point cloud segments to BIM components based on different types of features. (3) Conducting an experimental analysis to evaluate the performances of the developed matching approaches and effectiveness of the features used by these approaches. (4) Developing an approach that selects and combines point clouds captured at different times in order to reduce the file size of combined point cloud while maintaining the completeness of geometric information.