A repeatable mobile LiDAR job begins before anyone presses Start. The crew needs a named deliverable, a coordinate strategy, a safe walking or driving route, and a short list of checks that can be completed while a return visit is still cheap. That sounds basic. On a busy site, it is surprisingly easy to skip one of those decisions and discover the gap only after processing.
This field process is designed as a reusable run card. It applies to a handheld or backpack walk, a vehicle loop, and many robot or drone missions. Instrument-specific limits still govern. Artec lists Jet as a SLAM-based LiDAR system with handheld, backpack, vehicle, cage, telescopic-pole, robot, and drone deployment modes; the current product page also lists LAS, LAZ, PLY, and E57 point-cloud formats. Those capabilities create options, while the project requirements decide which options belong on a particular job.
1. Turn the requested output into acceptance tests
“Scan the building” gives a crew almost nothing to check. “Deliver a georeferenced E57 covering accessible mechanical rooms, with registered floor-control coordinates and a gap map” is actionable. Before mobilization, the project lead should write down the area of interest, excluded spaces, coordinate reference, required file formats, features that must be visible, and the tolerance used to accept the final product.
- Geometry: surfaces and objects that the client will measure, model, inspect, or compare.
- Position: local project coordinates, a published coordinate system, or a clearly documented arbitrary origin.
- Evidence: control report, checkpoint residuals, field log, coverage views, and processing notes.
- Packaging: full-resolution cloud, decimated viewing copy, trajectory if required, and a readable index.
The promised accuracy belongs here as a project requirement, never as a number copied from a brochure. Manufacturer accuracy, mapping accuracy, field geometry, control quality, capture path, and processing choices describe different parts of the error budget.
2. Walk the route without collecting
A five-minute reconnaissance can save an entire rescan. The operator looks for gates, elevator access, traffic crossings, active equipment, blank corridors, glass, water, dust, moving crowds, overhead obstructions, and areas where a closed loop can reconnect. The safety lead identifies exclusion zones and determines whether a spotter, traffic control, fall protection, mine authorization, or a certified remote pilot is required.
The planned path should contain recognizable geometry and deliberate overlap. Long featureless passages, repetitive racks, and open ground offer fewer constraints for SLAM. A return leg through varied, previously captured geometry gives the trajectory a useful opportunity to close. Where walking back is impossible, the plan should add cross ties or control observations rather than hope processing will solve the route.
3. Establish control and independent checks
Control connects the dataset to the required frame. Checkpoints test the result without participating in the adjustment. Mixing those roles makes a report look reassuring while revealing little about independent accuracy. The survey lead should select stable, recoverable locations, distribute them through the work, and record coordinate system, datum, units, survey method, point description, and expected precision.
Targets must be visible from useful approach angles and identifiable in the cloud. Their quantity and placement come from the project accuracy plan. A dense cluster beside the start offers weak evidence at the far end of a corridor. The current ASPRS positional accuracy standards emphasize declared accuracy classes, checkpoint quality, distribution, testing, and reporting; a project that invokes an ASPRS class should follow the applicable edition and addenda in full.
4. Prepare a capture packet
| Item | Recorded before Start | Reason |
|---|---|---|
| Job identity | Project, area, run ID, operator | Prevents mixed or orphaned files |
| Instrument state | Device ID, storage, power, time | Supports traceability |
| Environment | Weather, dust, lighting, traffic | Explains changed conditions |
| Control | Point IDs, units, reference frame | Avoids coordinate ambiguity |
| Route | Start, closure, cross ties, hazards | Makes coverage intentional |
File names should be assigned before capture. A useful pattern includes project, zone, date, run, and deployment mode. The operator also synchronizes clocks used by photos, field notes, or separate survey instruments. A one-page packet is more dependable than facts scattered across text messages.
5. Capture at a deliberate pace
The operator starts in a geometry-rich area, pauses long enough to confirm the run is recording, then follows the planned route smoothly. Rapid spins, abrupt platform motion, unnecessary stops, and people walking close to the sensor can complicate the cloud. Travel speed must remain within the instrument and platform guidance, and should slow around critical details, corners, elevation transitions, or reduced visibility.
Coverage deserves attention in three dimensions. Walking down the center of a room may show floors and walls while leaving pipe tops, beam seats, or equipment backs hidden. A second viewpoint is often worth more than another pass along the same line. The crew marks locked rooms and occlusions in the field log rather than allowing silence to imply complete coverage.
6. Verify before leaving the zone
Artec describes an on-site preview for checking Jet scans. A preview is most valuable when the crew uses a fixed inspection sequence: confirm the run duration and trajectory, orbit the cloud, clip through each critical elevation, inspect transitions, view target locations, and compare the coverage with the area-of-interest sketch. A colorful overview alone can hide thin data and shadowed surfaces.
At least one checkpoint should be tested early enough to reveal a units, datum, or target-identification mistake. Loop closure and control residuals should be reviewed against prewritten thresholds. If a check fails, the operator labels the run, captures a corrective tie or replacement run, and records the reason. Keeping failed runs can help diagnose the issue; presenting them as accepted data creates confusion.
7. Close out with a handoff that another person can follow
Back at the office, files are copied to two controlled locations and checked for readable size and naming. Processing should preserve a record of software version, coordinate transformation, control used, excluded observations, filtering, decimation, and export settings. Final QA compares independent checkpoints, searches for gaps at required features, checks units and orientation, and opens every promised format in its intended downstream application.
The repeatable process finishes with a field-to-office note: runs accepted, runs rejected, areas inaccessible, control file used, outstanding questions, and the person responsible for the next step. That humble note is one of the strongest tools in a mobile LiDAR workflow. It lets a different crew reproduce the capture next week and lets a reviewer understand what actually happened without reconstructing the day from filenames.