How to Validate the Accuracy of a Handheld SLAM Scanner: Check Points, Tolerances, and Test Methods
On this page
Validate Accuracy Against Project Tolerances and Independent Checks
Validate handheld simultaneous localization and mapping (SLAM) accuracy against written project tolerances using separate independent check points, traceable reference coordinates or dimensions, and a quantitative report. This guide explains how to establish check points, measure reference values, interpret mean absolute error (MAE), root mean square error (RMSE), and maximum deviation, and use SHARE PointClouds Studio V2.6 for 3D or elevation accuracy checks.
Step 1: Establish Independent Check Points
Check points are used to validate the deliverable, and the key principle is independence: they must not be used in control-point calibration, coordinate-transformation parameter estimation, or point-cloud registration. They are used only to check the completed result.
The number should be determined by project scale, applicable standards, and the acceptance party. Three to five points are only a starting point for a small pilot area—not a universally sufficient number for every project.
- Quantity: A small pilot area can start with 3–5 points. Increase the number for formal projects according to scale, standards, and acceptance requirements, and do not cluster them in one corner;
- Distribution: Cover near-, medium-, and long-range distance bands;
- Point selection: Use locations with clearly defined geometric features, such as wall corners, door-frame corners, and windowsill corners, so they can be located precisely in the point cloud;
- Marking: Use targets/reflective markers or fixed features that can be identified in the point cloud.

Figure 1: Quadrant targets distributed across different walls, columns, and spatial depths.
Step 2: Measure the Reference Values
Use a trustworthy instrument or method to measure the “reference values” between check points:
- Tape measure/laser distance meter: Suitable for spot-checking dimensions, but record the instrument specifications, calibration status, target surface, alignment method, and operating conditions;
- Total station or another traceable survey method: Suitable for establishing an independent 3D coordinate reference; select the appropriate grade according to project standards;
- Long-distance tape measurement: Control tension, level, temperature, and sag. A deviation observed in one published sample cannot be extrapolated into a fixed threshold;
- Repeated measurements: The number of repetitions and statistical method should be specified in the acceptance plan rather than mechanically applying a fixed count.
Step 3: Run the Check in SHARE PointClouds Studio V2.6
SHARE PointClouds Studio V2.6 provides two paths:
-
3D Accuracy Check: Import theCheckpoint Filecontaining reference coordinates, pick the corresponding named points one by one in the point cloud, runMatching Calculation, and useExport Reportto generate the 3D accuracy report; -
Elevation Accuracy Check: Import theCheckpoint File, setMatching Neighborhood(manual range: 0.1–1; default: 0.10), runMatching Calculation, and useExport Reportto generate the elevation accuracy report; - Dimension verification: For applications such as renovation measurements, distances between corresponding geometric features can also be measured in the point cloud and compared with trustworthy reference values;
- Operational quality: Use orthographic locking; endpoint, intersection, midpoint, and perpendicular-foot snapping; and the Ctrl magnifier to reduce picking differences.
Step 4: Calculate Error Metrics
A single deviation is not enough. Calculate at least three metrics:
| Metric | Meaning | Purpose |
|---|---|---|
| MAE | Mean of the absolute values of all deviations | Indicates the overall deviation level |
| RMSE (root mean square error) | Square root of the mean squared deviations | More sensitive to large deviations |
| Maximum deviation | Deviation at the worst point | Indicates the worst case |
Decision rule: Assess MAE, RMSE, maximum deviation, outlier distribution, and check point coverage together according to project standards and the acceptance party’s agreement. Do not declare compliance based only on an average or a single dimension.
Compare Results with Acceptance Tolerances (Building Examples)
| Standard | Permissible deviation | Notes |
|---|---|---|
| Example provisions in China’s GB 50204-2015 | Subject to the provisions adopted for the project | First confirm the measured object, provision version, and the acceptance party’s interpretation |
| Example provisions in Europe’s BS EN 13670:2009 | Subject to the provisions adopted for the project | Conclusions from a single sample cannot be extrapolated to every project |
| Example provisions in the US ACI 117-10(15) | Subject to the provisions adopted for the project | Verify the current version, component type, and measurement method |
These examples only illustrate how to verify standards and project provisions. They do not declare that any model complies with these standards in every application. The published SHARE SLAM S20 test involving 40 measurement pairs in a three-story villa is evidence only under its specific test conditions. Any citation should also include the method, sample, reference instrument, and data source.
How to Judge Whether an “Accuracy Figure” Is Credible
Before accepting any accuracy figure, confirm four things—all are essential:
- Test method: How was it measured?
- Test environment: In what setting and under what conditions?
- Testing party: Was the test conducted by the manufacturer or a third party?
- Sample scope: How many points were measured? An unqualified statement such as “1 cm accuracy” is not meaningful for comparison. See “How Should You Interpret SHARE SLAM S20 Accuracy and Reliability?”
Frequently Asked Questions
Q1: Can I judge accuracy from only one or two dimensions? Usually not. A small pilot area can start with 3–5 well-distributed check points. For formal projects, determine the quantity and distribution according to area, floor count, critical zones, coordinate requirements, and acceptance standards.
Q2: Is a tape measure or laser distance meter better as a reference? It depends on distance, target surface, traceability, and project requirements. For a tape measure, control tension, level, and sag. For a laser distance meter, control the reflective surface and incidence angle. If 3D coordinates or stricter control are required, use a reference method such as a total station that meets the project’s required grade.
Q3: How much error is acceptable? Refer to your acceptance tolerances. There is no universal pass value independent of the project. The acceptance party should confirm in writing the tolerances, statistical metrics, number and distribution of check points, outlier handling, and report format, and results should then be judged under those rules.
Q4: Can I validate SHARE SLAM accuracy myself? Yes. Use the method above at your own site. SHARE3DCAM also offers demonstrations and trial scans, allowing you to validate results on a real pilot project before making a decision.
Q5: Do I still need an accuracy check after using real-time kinematic (RTK) or post-processed kinematic (PPK) positioning? Yes. RTK status indicators, control-point calibration, and an external PPK tool-generated RTKLIB POS trajectory cannot replace an independent check. Also verify the source/target coordinate systems, elevations, units, and the X/Y/Z (east/north/elevation) order.
PROJECT SUPPORT

