How to Process Scans for Convergence Monitoring

Prev Next

This article explains how to use Aura's Change Detection and Convergence Monitoring workflow to compare two Hovermap scans of the same area and visualize changes between them. The result is a point cloud with point-to-mesh distance attributes and a mesh file generated from the reference scan.

Aura's Change Detection and Convergence Monitoring solution integrates the rapid data capture capabilities of Hovermap with Aura's processing and analysis to help manage and monitor excavation projects. Tailored for underground environments and enclosed spaces, the solution captures current excavation profiles with Hovermap, then processes, aligns, and visualizes changes in Aura.

The following video shows how to use Aura to compare two Hovermap scans of the same area.

What you will need

  • Operating system: Windows 10

  • Aura: Version 1.8.1 or higher

  • Entitlement: A valid SLAM and Convergence Monitoring license

  • Point cloud data: 2 Hovermap scans of the same area, with the original .bag files for both scans

This solution is designed and optimized for use in underground mining (drives and tunnels) and other internal enclosed spaces. It may not be suitable for other environments.

Scans can be captured by walking, driving, or flying.

Before running the Convergence Monitoring workflow, both scans must be processed using SLAM.

How it works

The basic process for change detection and convergence monitoring involves taking the reference scan, creating a mesh of that reference scan, and then measuring the distance from the points in the second scan (of the same area) to the mesh.

Procedure

Step 1: Configure the convergence monitoring job

  1. Open Aura. Ensure an active SLAM and Convergence monitoring license is present.

  2. In the Process tab, click Process Scan.

  3. In the Configure New Scan Job panel, select the Convergence monitoring workflow.

  4. Click Add Datasets.

  5. In the dialog box that displays, click Add, then browse for the point clouds to use.

Ensure the point clouds have been pre-processed in the latest version of Aura.

  1. In the Alignment Preview File column, select the scan file to use for alignment.

Convergence Monitoring is performed on the full point cloud regardless of the selected Alignment Preview File. Use only the subsampled files to lower the computational resources needed, especially for large datasets. The reduced set of points on a subsampled file also helps focus on key features for more accurate alignment.

  1. Select the reference scan file. The reference scan provides a fixed frame of reference for aligning the other point cloud, establishing a coordinate system that other scans can be transformed into.

  2. Once the point clouds have been added and the reference scan selected, click Save.

  3. In the Location field, enter the preferred name for the output folder and select the output folder location.

  4. Select the profile to use.

    Configuration settings for a convergence monitoring scan job in a software interface.

  5. Click Processing Settings to configure the convergence parameters. See the Processing settings section below for detailed information.

Skip this step to process using the default settings.

  1. Click Next.

Step 2: Review and manually align the datasets

  1. Click Load Scans to start the review. All scans load at the same time.

If the incorrect dataset was selected, click Edit Point Clouds Selected to load the correct ones.

  1. Follow these guidelines to roughly align the second scan with the reference scan:

    1. Ensure the second scan is selected in the Visualize tab under the Point Clouds menu.

    2. Use the Transform and Rotate tools to align the point clouds.

When using the Rotate tool, the following shortcut keys change the pivot point of the rotation:

  • CTRL+Shift+C: Changes the origin point to the newly selected point. Click an area on the second scan to select a new origin point.

  • Ctrl+Shift+Z: Resets the origin point back to the point cloud's origin.

  1. Once the scans have been roughly aligned, click Start.

Step 3: Start processing

  1. After clicking Start, a progress bar shows how far along the processing job is. The elapsed time of the processing job is shown to the right.

    The directory file path below the progress bar provides a way to identify the dataset source. This is useful if multiple jobs are processed simultaneously with the same output folder name. Copying the file path and pasting it into the file explorer allows access to the completed files without waiting for the processing job to be completed.

The Retry button becomes available when a failure occurs during processing. Click to attempt to process the current job from the last successful stage.

Step 4: Inspect the final output

Once processing is complete, two output files are generated:

Output

Description

.LAZ

Point cloud with point-to-mesh distance attributes.

.PLY

Mesh file created from the reference scan.

  1. Click View to display the generated .LAZ file in the Viewport.

    Convergence monitoring completed with file paths and sizes displayed for review.

Currently, aside from viewing the .PLY file, there are no options to interact with a Mesh file.

  1. In the Context Panel on the right, click Color Scale, then select Distance from the drop-down menu. The Step Change Detection - Rainbow color gradient is automatically selected.

    3D visualization of convergence monitoring data with various settings and options displayed.

The colors and distance values can be edited within the Scalar Filter. The distance points included in the final output file (.LAZ) can also be adjusted:

  • ADD: Click the + icon, then select a color and specify the Location in meters.

  • EDIT: Click the square color box, then click the Pencil icon.

  • DELETE: Click the square color box, then click the Trash icon.

User interface for scalar gradient and filter settings in data analysis software.

  1. Use the near clip plane (by holding the ALT key while using the mouse scroll wheel in Perspective mode) to view and navigate inside the tunnel.

Pieces of infrastructure (such as pipes, rock bolts, fans) may show changes due to how the meshing process removes and smooths out those features on the reference scan. These changes can typically be ignored.

Outcome

The two scans have been processed and aligned, and the resulting .LAZ point cloud is available for viewing with point-to-mesh distance attributes color-coded by the Step Change Detection - Rainbow gradient. The corresponding .PLY mesh file generated from the reference scan is also available in the output folder.

Additional information

Processing settings

Alignment settings can be changed via the Merge and Meshes tabs in Processing Settings.

When the settings on a built-in processing profile are updated in Aura, a temporary Custom profile is created, which can be used in any of the workflows for the current session. Save this custom profile to save time setting up processing jobs for common or known environments. Once saved, it becomes available for selection in the Profiles dropdown list.

Custom profiles that are not saved are automatically removed when the application is closed.

Merge tab

Field

Description

Voxels

  • Voxels Size: The lowest size of voxel used to generate surfels in SLAM. Use this inside smooth tunnels or bores, as most of the information within the points will be in the subtle variations in the surface direction that occur in relatively small dimensions. Increasing this number can significantly increase the processing time. Default setting: 0.4 m.

  • Voxel Levels: The number of levels used to generate surfels in SLAM. Each level is twice the size of the previous level. Use this inside tunnels or bores, as most of the information within the points will be in subtle variations in the surface direction that occur in relatively small dimensions. Default setting: 5.

  • Voxel Minimum Points: The minimum number of points in a voxel to use for SLAM. Use this to reduce the impact of noisy data on SLAM, or to ensure that features with a low number of points are included in environments with few geometric features. Default setting: 8.

Matching

  • Number of Matches: Number of surfels to look for matches during SLAM.

  • Max Distance: Maximum distance (in voxels) to search for surfel matches.

Meshes tab

Field

Description

Size of triangles (mm)

The level of detail that makes up the 3D mesh. A smaller triangle size results in a more detailed mesh.

Point filtering (m)

  • Range: Filters the points outside the specified range from the scanner.

  • Denoise SOR: Removes outlier points that are less likely to be real.

    • Nearest neighbor: The number of point neighbors used for data analysis. A smaller value results in quicker processing and less statistical certainty.

    • Log scale: The threshold for filtering. A smaller value results in more aggressive filtering. 8 = conservative; 6 = default; 4 = aggressive.

  • Adaptive SOR: Removes the points that seem fuzzier than nearby points.

    • Nearest neighbor: The number of point neighbors used for data analysis. A smaller value results in quicker processing and less statistical certainty.

    • Alpha: The threshold for filtering. A larger value results in more aggressive filtering. 0.05 = conservative; 0.10 = default; 0.50 = aggressive.