Automatic modeling of prefabricated components with laser-scanned data for virtual trial assembly

Virtual trial assembly (VTA) simulates the preassembly by verifying the correspondence of as-built and as-designed prefabricated component (PC) models in virtual reality. Notably, obtaining accurate as-built PC models ensures the success of VTA. To accomplish it, current methods require matching the scanned PC data with the as-designed model and manually modifying the model to approximate the actual dimensions, which is inefficient and leads to subjective errors. Therefore, it is desirable to develop an automatic modeling approach to obtain reliable as-built PC models for further improving the efficiency of VTA. However, to the best of our knowledge, few studies have been done on the automatic modeling of as-built PC models in VTA procedure. To this end, based on the prior knowledge of as-designed model, this study proposes an approach to automatically reconstruct the primary structure via plane fitting and structural feature point calculation for geometric modeling. Then, two adaptive methods for inner boundary detection and round-hole estimation are developed for the reconstruction of detailed structure. Finally, as-built model visualization is realized based on the data extraction and modeling development of a building information modeling software. To verify the feasibility of the proposed approach, a validation experiment was conducted on a full-scale precast concrete stair using two scanning resolutions (1/5 and 1/8). Experiment results show that the estimation of the dimensions consists of multifaceted corners that are stable and not affected by scanning resolution. Low scan resolution (1/8) may increase estimation errors of the edge, which has a single-sided turning point. The experiment results also confirm the effectiveness of the proposed approach for the automatic model reconstruction of as-built PCs.