Construction monitoring system based on point cloud analysis
Early detection of deviations from the project and reduction of construction control costs
- Enterprise software
- Qt Framework
- Point Cloud Library
To monitor construction, laser scanning technology is often used when the structure is scanned with a 3D scanner, and the resulting point cloud is compared with the original 3D model of the structure.
To match the cloud with the model, the engineer looks for each element of the structure and outlines the points that belong to it. This process takes up to one week. The quality and speed of comparison directly depend on the specialist's qualifications, and any mistake can result in a change in the timing and cost of construction or even a substandard structure.
Rubius, upon request from Gazprom Neft, has developed a program that compares a cloud to a model in 30 seconds and finds inconsistencies that are invisible to the human eye. It reduces the cost of construction monitoring, helps to detect deviations from original designs timely, adjusts the work plan, and minimizes the risks of non-compliance with construction deadlines and budget. The system can also be used to inspect structures or assess damage after accidents and emergencies.
- processes the cloud – selects the corresponding parts of the model, removes noise and downgrades to a given density,
- aligns the model and the cloud by a user-defined structural element or by 4 control points selected by the system at the intersection of horizontal and vertical planes (for example, the ground and walls of a building),
- segments the cloud – recognizes structural elements in it (beams, pipes, wires, etc.) and classifies them based on data from the library of typical parts,
- compares the point cloud with the 3D model and looks for inconsistencies,
- evaluates inconsistencies based on the specified parameters (dimensions, position in space, accuracy of surfaces),
- generates tables with calculations of deviations of the structure from the project,
- supports STL, FBX, PLY for 3D models, PCD, PLY, and LAS for point clouds.