LiDAR boresight calibration aligns the laser scanner's orientation with its navigation reference. A small angular error becomes a larger position error at longer range, so a point cloud can look sharp nearby while opposite survey strips separate farther away. Diagnose mounting, translation and timing separately before adjusting the calibration.
Table of Contents
Align two coordinate frames
A LiDAR scanner measures returns in its own frame. A global navigation satellite system and inertial navigation system, or GNSS/INS, describe the platform trajectory and attitude in another. Georeferencing combines the two through a rigid transformation and a time association.
The angular relationship is the boresight; the translation between sensor origins is the lever arm. Treating both as a single “GPS accuracy” number makes faults difficult to diagnose. A translation error shifts the origin. An angular error rotates the measured direction and becomes increasingly important at longer range.
OxTS’s boresight calibration instructions, updated in May 2026, distinguish angular alignment from optional linear refinement. They also make navigation quality a prerequisite. This is a specific implementation example, rather than a universal procedure for every LiDAR package.
How large can a small angle become?
For a small angular error θ in radians and target range R, the transverse displacement is approximately e = Rθ. This geometric estimate isolates one error source; it is not a complete survey accuracy prediction.
An illustrative 0.1° error equals approximately 0.001745 radians. At 20 m, the offset is about 35 mm. At 100 m, it is about 175 mm. Which coordinate receives that error depends on the axis and viewing geometry.
A short-range calibration scene may therefore fail to reveal a problem that matters in a higher-altitude survey. Use geometry representative of the intended measurement, while following the calibration software’s supported collection procedure.
Separate the signatures before fitting parameters
Different faults can produce similar-looking clouds. Use multiple directions, distances and surfaces to distinguish them.
| Observed pattern | Candidate explanation | Useful independent check |
|---|---|---|
| Separation grows with range | Angular alignment error | Compare near and far surfaces |
| Roughly consistent displacement | Lever arm or reference offset | Inspect measured sensor origins |
| Difference changes with speed | Time association or trajectory issue | Compare timestamps and speed dependence |
| Whole project has height bias | Datum or control mismatch | Check independent vertical references |
| One strip is locally distorted | Navigation degradation | Review trajectory quality for that interval |
The table is a diagnostic guide, not an automatic classifier. Several faults can coexist. Fitting a flexible correction to poor navigation may reduce a residual without identifying the physical problem.
The vertical datum guide addresses project-wide height references. Camera internal parameters belong to photogrammetry calibration, which is a different calibration task.
Deliver a calibration that can be reused
Keep the sensor identities, mount configuration, coordinate-axis convention, units, trajectory-processing version and calibration parameters together. A number called “roll offset” is ambiguous without direction and frame definitions.
Use one dataset to estimate parameters and an independent set to inspect the result. The second set should include the operating ranges and surfaces that matter to the project. Compare both local strip agreement and independent control; improvement in one does not establish the other.
A practical verification set includes a short baseline at the project’s typical altitude, a longer line at its maximum working range, and a repeat pass in the opposite direction. If the project uses control points, keep a subset out of the adjustment. Record the residuals by direction and range; a single RMS value can hide a systematic angular pattern.
When specifying an industrial UAV mapping payload, assign responsibility for mount measurement, synchronization, calibration and final output. For agricultural mapping, terrain consistency and repeatable reference frames matter when comparing surveys over time. Include the configuration record with the technical handover resources.
The reusable deliverable is the transformation and its supported configuration, accompanied by a verification result. A visually attractive point cloud alone cannot show which mounting, timing and reference assumptions produced it—and cannot tell the next survey team whether the calibration still applies.
FAQs
Does boresight calibration replace ground checkpoints?
No. Alignment can improve agreement between strips while the whole survey still carries a position or height bias. Independent checkpoints assess the delivered survey in its required reference frame.
When should a LiDAR boresight calibration be repeated?
Repeat the affected calibration after a mounting change, impact, sensor replacement or unexplained strip mismatch. Preserve the configuration and compare a verification dataset before reusing old offsets.
Can a timing error look like a mounting error?
Yes. Motion during a time offset creates a position discrepancy and can distort strip agreement. Analyze synchronization and trajectory quality before fitting additional mounting parameters.
What should be recorded with the calibration result?
Sensor identities, mount configuration, axis convention, units, trajectory-processing version and the verification residual. Without those fields, the same offset cannot be applied safely to another aircraft or payload.
