How Can Teams Validate SHARE SLAM in Complex Spaces? Glass Curtain Walls, Long Corridors, and Weak-Feature Areas

For Complex Spaces, “Can It Scan?” Is Not the Only Question

For simultaneous localization and mapping (SLAM) workflows, long corridors, glass curtain walls, or indoor-outdoor coordinate continuity, evaluate coverage, trajectory, coordinates, and acceptance separately. SHARE3DCAM’s SHARE SLAM S20, SHARE SLAM S20 SE, SHARE SLAM S100, and the SHARE C10 Series should be validated at the actual site. The model, route, and control plan must be determined together; results apply only to the recorded conditions.

Figure 1: Complex-space sample validation proceeds through risk identification, route and control planning, sample acquisition, office inspection, independent validation, and solution adjustment.

Six High-Risk Environments and How to Validate Them

How to Connect Segmented Acquisitions

When the spatial relationships between segmented point clouds must be solved, use manual corresponding-point registration. Specify the stationary reference point cloud and the point cloud to be transformed, then pick the same features in both viewports, such as wall corners, building corners, fixed ground points, or targets. Corresponding points should cover the full area and should not be clustered or collinear. After previewing overlap and error, save the result as original_name_registered.las; retain the source point cloud. Merging is only for point clouds that are already aligned and cannot replace registration. If corresponding points are difficult to identify reliably in long corridors or repetitive spaces, add stable features or control during acquisition planning rather than forcing matches during office processing.

How to Handle Indoor-Outdoor Coordinates and External Post-Processed Trajectories

When GNSS conditions are available, the device’s built-in real-time kinematic (RTK) trajectory can be used. Alternatively, an external post-processed kinematic (PPK) tool can process rover, base-station, and navigation data to generate an RTKLIB POS trajectory file, which is then selected only as the Trajectory Source in GNSS Fusion. Before and after import, inspect trajectory continuity, gaps, fixed-solution coverage, source and target coordinate systems, and height offsets. PPK is not a shortcut that eliminates external processing and quality validation. To transform an existing point cloud into engineering coordinates, import control points and match them individually. Verify X/Y/Z against easting, northing, and elevation, respectively, and use at least 3 valid control points to calculate transformation parameters. Alternatively, enter confirmed XYZ translation, XYZ rotation, and scale parameters. Remove high-error points and recalculate, then save the result as original_name_converted.las.

Quality Checks Must Be Independent

Control points are used to solve the transformation, while check points verify the result; the two should remain separate.

SHARE PointClouds Studio can perform:

• 3D Accuracy Check: Import the independent Checkpoint File, manually select corresponding positions in the point cloud, run Matching Calculation, and use Export Report to generate the 3D accuracy report.
• Elevation Accuracy Check: Set Matching Neighborhood, run Matching Calculation, and use Export Report to generate the elevation accuracy report. Point-cloud comparison is suitable for visual comparison of changes after coordinates have been unified, but it does not produce quantitative deviations and cannot serve as the sole basis for acceptance. 3D Gaussian splatting (3DGS) is suitable for photorealistic viewing, not for determining dimensions, coordinates, or accuracy.

When to Adjust or Change the Solution

If real-project testing shows persistent gaps in critical areas, a trajectory or coordinates that cannot be reviewed reliably, independent checks that do not meet project tolerances, or supplemental scanning and manual-processing costs beyond what the project can bear, adjust the route, add control, segment the acquisition, or evaluate combined solutions such as TLS, aerial surveying, and specialized surveying services. Conclusions should include the site, weather, route, device configuration, and acceptance criteria and should not be generalized from a single demonstration to every project.

Frequently Asked Questions

Q: If a long-corridor point cloud shows level shifts or offsets, what should I check first? First check whether the route created loops and cross-constraints, then use linked navigation to inspect the trajectory, photos, and anomalous locations. Also verify the control-point distribution, corresponding points used for segmented registration, and independent check results.

Q: Must a continuous indoor-outdoor project use RTK? Not necessarily. Based on the target coordinates and on-site GNSS conditions, choose among RTK, an external PPK tool-generated RTKLIB POS trajectory file, control-point coordinate transformation, or segmented registration, and verify the result with independent check points.

Q: Can the software generate a PPK trajectory directly? The workflow in the manual first creates an RTKLIB POS in an external PPK tool, which is then selected only as the GNSS Fusion Trajectory Source. The trajectory and coordinate configuration must also be checked.

Q: Can 3DGS fill in glass and black surfaces as survey data? No. 3DGS is for photorealistic viewing and does not replace point-cloud coverage or survey validation. Address missing areas by adjusting the route, performing supplemental scans, or using another solution.

Q: If multiple point-cloud segments are already in the same project, does that mean registration is complete? No. Neither project management nor point-cloud merging solves spatial relationships. An _registered.las deliverable is generated only after manual corresponding-point registration and error review.