Systems and methods for automated spatial change detection and control of buildings and construction sites using three-dimensional laser scanning data

1. A method for automated spatial change detection for a construction site, the method comprising:

obtaining an as-designed model of the construction site;

receiving three-dimensional scan data of the construction site, the construction site having a network of a plurality of elements;

generating an as-built model of the construction site from the three-dimensional scan data, the as-designed model and the as-built model each including a plurality of three-dimensional objects corresponding to the plurality of elements;

aligning the as-designed model with the as-built model;

generating a relational graph for each of the as-designed model and the as-built model by detecting one or more local attributes of the three-dimensional objects present in both the as-designed model and the as-built model and computing a spatial relationship between the three-dimensional objects;

determining a deviation between the as-built model and the as-designed model in the relational graphs for each of the three-dimensional objects;

eliminating any of the three-dimensional objects with the deviation below a threshold using nearest neighbor searching;

isolating any of the three-dimensional objects with the deviation above the threshold using context matching;

generating a local spatial context for each of the isolated three-dimensional objects; and

associating any unmatched objects in the as-built model with corresponding objects in the as-designed model using the local spatial context.

2. The method of claim 1 , wherein the network is a ductworks with the plurality elements each being a cylindrical duct.

3. The method of claim 1 , wherein the as-designed model is aligned with the as-built model using a constrained iterative closest point registration within a maximum distance value, any portions of the as-designed model, and the as-built model outside the maximum distance value being removed.

4. The method of claim 3 , wherein the maximum distance value is 0.3 m.

5. The method of claim 1 , wherein the relational graphs are each generated by automatically extracting axes of the three-dimensional objects and generating line representations using the axes, the relational graphs each including a relative orientation and position and a spatial context of each of the line representations.

6. The method of claim 5 , wherein the line representations are each generated by extracting a line segment using a principle component analysis and calculating a center point of the line segment.

7. The method of claim 1 , wherein the threshold is 0.15 m.

8. The method of claim 1 , wherein the nearest neighbor searching identifies matched line representations and unmatched line representations of the three-dimensional objects between the as-built model and the as-designed model.

9. The method of claim 8 , wherein the isolated three-dimensional objects include the unmatched line representations.

10. A method for automated spatial change detection for a construction site, the method comprising:

obtaining an as-designed model of the construction site;

receiving three-dimensional scan data of the construction site, the construction site having a network of a plurality of elements;

generating an as-built model of the construction site from the three-dimensional scan data, the as-designed model and the as-built model each including a plurality of three-dimensional objects corresponding to the plurality of elements;

generating a first relational graph for the as-designed model;

generating a second relational graph for the as-built model;

eliminating any subnetworks of one or more matched objects between the as-built model and the as-designed model, the one or more matched objects having a deviation between the first relational graph and the second relational graph below a threshold;

extracting any subnetworks of one or more unmatched objects between the as-built model and the as-designed model, the one or more unmatched objects having a deviation between the first relational graph and the second relational graph below the threshold;

separating the subnetworks of one or more unmatched objects into a first category and a second category;

filling the one or more unmatched objects in the first category between a subset of the one or more matched objects using connection tracking; and

matching the one or more unmatched objects in the second category using local spatial context.

11. The method of claim 10 , wherein the network is a ductworks with the plurality of elements each being a cylindrical duct.

12. The method of claim 10 , wherein the threshold is 0.15 m.

13. The method of claim 10 , wherein the first and second relational graphs are each generated by automatically extracting axes of the three-dimensional objects and generating line representations using the axes, the first and second relational graphs each including a relative orientation and position and a spatial context of each of the line representations.

14. The method of claim 10 , wherein the as-built model is aligned with the as-designed model.

15. The method of claim 14 , wherein the as-designed model is aligned with the as-built model using a constrained iterative closest point registration within a maximum distance value, any portions of the as-designed model and the as-built model outside the maximum distance value being removed.

16. The method of claim 10 , wherein the first category includes the subnetworks having three or fewer connected three-dimensional objects and the second category includes the subnetworks having more than three connected three-dimensional objects.

17. A method for automated spatial change detection for a construction site, the method comprising:

obtaining an as-designed model and an as-built model of the construction site having a network of a plurality of elements, the as-designed model and the as-built model each including a plurality of three-dimensional objects corresponding to the plurality of elements, the three-dimensional objects segmented through registration of the as-built model with the as-designed model within a maximum distance value;

generating a first relational graph for the as-designed model;

generating a second relational graph for the as-built model;

isolating one or more subnetworks of the three-dimensional objects having a deviation between the first relational graph and the second relational graph above a threshold; and

matching each of the three-dimensional objects in the one or more isolated subnetworks using at least one of connection tracking or local spatial context.

18. The method of claim 17 , wherein the one or more subnetworks are isolated based on nearest neighbor searching and constraint propagation.