Robotic Assembly

As a studycase for the robotic assembly, we fabricated a spatial structure with a total of thirteen elements of square section connected through bolted face lap joints. The setup included two ABB GoFa CRB 15000-5 and a human.

Download .gh file

Steps

1. Input the data

First things first, let’s import your cleaned scan and corresponding polysurface model in Rhino. For this tutorial, we assume that the two are already aligned to each other as the location of the robotically assembled structure is known.

2. Build the DFAssembly

Here we convert the model of our structure into the internal datatype of diffcheck, DFAssembly.

DF’s components:

• DFAssebmly

3.Preparation of CAD and scan data

We get the pointcloud for the CAD, reduce them to decrease the computation time and compute the normals which is necessary for the next step.

Hint

We assume that the tranformation between the scan and the CAD model is known. If this is not the case, refer to the Registration step of the Manual Assembly

DF’s components:

• DFBrepToCloud

• DFCloudVoxelDownsample

• DFCloudNormalEstimator

4. Segmentation of the scan

We segment the scan to isolate the parts of the raw point cloud of the scan that correspond to each beam of the assembly.

DF’s components:

• DFCloudNormalSegmentator

• DFRemoveStatisticalOutliers

• DFJointSegmentator

• DFColorizeCloud

5. Error computation

At this point we can compute the error between the CAD model and the scan. The error is computed as the distance between the closest point on the CAD model and the scan. The current DF’s output metrics are:

• distance : the distance between the closest point on the CAD model and the scan

• mean : the mean distance between the closest point on the CAD model and the scan

• max_deviation : the maximum distance between the closest point on the CAD model and the scan

• min_deviation : the minimum distance between the closest point on the CAD model and the scan

• std_deviation : the standard deviation of the distance between the closest point on the CAD model and the scan

DF’s components:

• DFCloudMeshDistance

6. Error Visulization

DF allows you to quickly visualize the errors in the Rhino viewport. The color of the points represents the distance between the CAD model and the scan. The color scale can be adjusted to better visualize the error. We also provide a graph that shows the distribution of the errors.

View of the visualization of the mean error for the analysed clouds on the CAD model itself.

View of the graph of the corresponding distribution of the mean error directly in Rhino.

DF’s components:

• DFVisualizationSettings

• DFVisualization

7. Export the results

The results can be also exported in a CSV file for further analysis or documentation.

The CSV contains the values per element..

Beam ID	Min Deviation	Max Deviation	Std Deviation	RMSE	Mean
0	0	0.024	0.0029	0.0045	0.0034
1	0	0.0239	0.0034	0.0057	0.0046
2	0	0.0316	0.0036	0.0056	0.0042
3	0	0.024	0.0069	0.0088	0.0055
4	0	0.0117	0.0029	0.0053	0.0045
5	0	0.0242	0.0045	0.0067	0.0049
…	…	…	…	…	…

DF’s components:

• DFCsvExporter