Generating Topologically Consistent BIM Models of Utility Tunnels from Point Clouds

Area of Science:

• Geomatics Engineering
• Civil Engineering
• Computer Vision

Background:

• Urban underground space development is crucial for addressing urban challenges.
• Utility tunnels are vital infrastructure components requiring accurate investigation.
• 3D laser scanning in utility tunnels faces challenges like noise and occlusion, impacting data quality and BIM topology.

Purpose of the Study:

• To propose a hierarchical point cloud segmentation method for utility tunnel data.
• To develop a robust topology reconstruction method for Building Information Modeling (BIM) objects in utility tunnels.
• To address inaccuracies in geometric element extraction and ensure the topological consistency of reconstructed BIM models.

Main Methods:

• Hierarchical segmentation of point clouds into facades, planes, and pipelines.
• An improved mean-shift algorithm for extracting wall line features.
• A local symmetry-based medial axis extraction algorithm for pipeline identification.
• A neighbor-searching approach to reconstruct topologically consistent BIM models based on wall and pipeline centerlines.

Main Results:

• Successfully segmented point clouds and extracted key features within utility tunnels.
• Reconstructed 24 BIM wall objects and 12 pipelines from a real-world dataset.
• Demonstrated the efficiency and effectiveness of the proposed methods in creating topologically consistent 3D utility tunnel models.

Conclusions:

• The proposed hierarchical segmentation and topology reconstruction methods significantly improve the accuracy and consistency of BIM models for utility tunnels.
• This approach overcomes limitations of traditional 3D laser scanning in complex underground environments.
• The validated method enhances the reliability of BIM for urban underground infrastructure management and planning.