This article explains how to scan a long, linear route, such as a road or a tunnel.
Long, linear scans are the most difficult conditions for mobile LiDAR. These environments usually have few features, and the scanner uses features to remove accumulated drift. Poor capture technique can result in SLAM slip or SLAM drift.
What you will need
An Emesent scanner set for the scan platform in use (handheld, backpack, vehicle, or drone).
Emesent Commander, to monitor the live point cloud during the scan.
Aura, to process the scan data after the scan.
A georeferencing method, if the point cloud must sit in real-world coordinates.
This article covers only what is different about a long, linear route. Complete the standard planning and setup first:
Choose a georeferencing method
RTK and PPK both correct SLAM drift and put the point cloud in real-world coordinates, so either is sufficient for most linear scans above ground. Use GCPs instead when GNSS is not available, or when the accuracy required is higher than RTK or PPK can give.
Use GCPs when:
The scan must align to a survey or a design model.
Multiple scans must sit in a shared coordinate frame.
The project requires survey-grade or repeatable accuracy.
The site prevents GNSS use, but global positioning is still required.
For all four methods and how they compare, see Georeferencing.
Choose the scan path
Use two passes along the route, one in each direction, and offset the two passes from each other. Start and finish the scan in the same location, so that the loop closes.
The best path is different for each scenario. Design the path around the access available, the area to capture, and the need to close loops. For the general technique, see Scanning techniques.
The two examples that follow are both roads. In the first, a drone has full access above the route. In the second, the route can only be driven.
Obey the local traffic regulations and the site rules. Do not move between lanes on a live road, and do not cross a live traffic lane on foot.
Free access across the route
This is the ideal condition for loop closure. A drone flight or a closed road gives access across the full width. Move in a zig-zag across that full width. Keep 30 m to 40 m between each crossing of the center of the route. Frequent crossings close loops often during the scan, and give complete coverage of the capture area.
Scan path where the entire area is available
Restricted access across the route
This is the second choice. Capture what the available access permits. A live road does not give access across the full width. Keep to one side of the route for the outbound pass. Use the other side for the return pass. A vehicle stays in its lane. On foot, use the shoulder or the verge.
Scan path where access across the route is restricted
Above ground and GPS-denied routes
The steps are the same for every long, linear route. The table that follows gives what changes between the two conditions.
Consideration | Above ground | GPS-denied |
|---|---|---|
Typical routes | Roadways, highways, rail corridors, and service corridors. | Tunnels, underpasses, and underground access drives. |
Georeferencing | RTK or PPK. Use GCPs for survey-grade accuracy. | GCPs. |
Features to keep in view | Signs, barriers, and roadside structures. | Portals, intersections, and variation in the walls. |
Surfaces to be careful of | Uniform pavement, and open ground beside the route. | Long smooth walls, and reflective or wet surfaces. |
Coverage to confirm | The road surface, the shoulders, and the intersections. | The walls, the ceiling, and the floor. |
Access | Agree the scan windows, the permits, and the traffic restrictions. | Record the lighting, the moisture, and the access limits. |
Procedure
Step 1: Start the scan
Complete the pre-mission setup for the platform and the georeferencing method in use. This includes the GCP targets, and the RTK or PPK equipment. See Prepare for a mission.
Select a start position at the halfway point of the route, where the environment is feature-rich. If the halfway point has too few features, offset the start position along the route.
Start the scan, then follow the planned path at the speed recommended for the platform in use.
Keep distinct features within 40 m of the scanner. Where possible, stay 5 m to 15 m from the surfaces of interest for better point density.
Step 2: Follow the scan path
Complete the outbound pass to the first end of the route. Keep the same side, or the same crossing interval, for the full length of the pass.
Turn around at each end at a steady pace.
Complete the return pass, offset from the outbound pass.
A slow turn gives the scanner time to see the new features and the old features together. The two passes see the route from opposite directions and from different positions. This gives the scanner loop closures, which correct drift and fill the gaps in the data.
Step 3: End the scan
Move the scanner back to the start position, then continue a short distance past it. This closes the loop at the start.
Examine the live point cloud in Commander and confirm that the coverage is complete.
Stop the scan, then offload the scan data.
Step 4: Process the scan data
The parameters in the table are correct for most linear scans, but the result depends on the environment. A change beyond the recommended parameters increases the processing time by a large amount.
Process the scan data in Aura with the settings recommended for the environment type. See Processing scan data and Aura scan environments.
If slip, drift, or accuracy problems occur, first change the Local Mapping parameters to the values that follow. For the other settings, see Processing troubleshooting.
Section | Parameter | Value |
|---|---|---|
Local Mapping | Sliding size in seconds | 15 |
Local Mapping | Sliding shift in seconds | 0.6 |
Outcome
The route is scanned, and the scan data is processed in Aura.
Additional information
For assistance, contact the regional Emesent partner or the Emesent Client Support team.
