Building Information Modeling Object Accuracy Analysis Based on Point Cloud Using an Unmanned Aerial Vehicle (UAV)

Building Information Modeling has an important role in the construction industry. One of the important jobs in the BIM process is 3-dimensional modeling which is applied at the planning stage, during construction and post-construction. One of the challenges in 3-dimensional modeling is at the post-construction stage, because the accuracy of the 3-dimensional model size must be close to existing. The 3-dimensional modeling method using as-built drawings is considered less accurate because the documents may not be current. One of the most accurate and actual data acquisition methods for existing conditions is digital scanning using a laser scanner which produces point cloud data. Point cloud is used as the basis for 3-dimensional modeling. Point clouds can also be generated from photogrammetry methods based on aerial photos using conventional Unmanned Aerial Vehicles (UAV). The problem faced is that the price of laser scanners is much more expensive than conventional UAVs. Meanwhile, 3-dimensional point cloud-based modeling from photogrammetry with UAVs is not yet commonly applied because it is considered less accurate than laser scanners. If the level of accuracy of point cloud-based 3-dimensional modeling of UAVs can be known, then it can be considered the feasibility of implementing it as a much cheaper alternative for data acquisition. The accuracy of indicator for this research uses Root Mean Square Error (RMSE) analysis. The research results show that by utilizing the point cloud from the UAV, the average accuracy level of the 3-dimensional model of cubic square elements reach 96%. Meanwhile, the accuracy of curved or circular elements is only around 56%. This level of accuracy shows that conventional UAVs can be an alternative tool that is easy to obtain, economical and feasible for BIM modeling data acquisition.