Acquiring Semantic Information of Precast Concrete Pipe Using Geometric Feature Extraction from Mobile Laser Scanning Point Cloud

Abstract

In a construction project, the need to conduct effective and efficient material management is urgent since it takes up to 60% of the project budget. Most of the material tracking uses tag-based technology by attaching RFID, GPS, or UWB to the materials. This method is found to be effective in tracking the material in construction projects. However, there is still a manual job of putting the tag into each of the materials and problems related to the signal quality and infrastructure requirements. As an alternative way to do material tracking, point cloud processing can be used. This paper aims to develop an efficient approach to accurately detect concrete pipe precast material, calculate its numbers, and produce valuable data for material inventory management. The data was taken by a mobile laser scanner of a site-specific infrastructure project in Sydney. The resulting data was sparse and occluded because it was taken only from one side to replicate the actual scanning process in a construction project. The automated process has been conducted by matching the material based on its geometric feature of the 3D material model. The proposed approach can provide some spatial information such as location (x, y, z global coordinate), orientation of each material, and the number of materials. The result can detect up to 78.5% of the material. The difference between actual and predicted global coordinates is 0.75m which is acceptable for material location in a large infrastructure project. This data can be reconstructed in a 3D detail of the site project in the Building Information Modeling (BIM) platform in its actual location. The implementation of this method serves as an initial stage toward achieving synchronization between the physical construction and its corresponding digital twin in the field of construction.