As an essential foundational dataset in highway survey and design, digital elevation models (DEMs) have traditionally been acquired through labor-intensive methods such as GNSS RTK and total station measurements. Highway projects often span long distances and traverse complex terrain, making conventional data collection both time-consuming and costly. In recent years, China drone LiDAR technology has emerged as a transformative solution, delivering high-precision point clouds and enabling rapid production of DEMs. In this paper, I present my practical experience applying China drone LiDAR to a real highway survey and design project, demonstrating that the technology not only meets stringent accuracy standards but also dramatically improves work efficiency by a factor of three or more.
1. Introduction
The rapid expansion of highway infrastructure in China demands ever-shorter design cycles and higher-quality topographic data. Digital elevation models are the backbone of route alignment, earthwork estimation, and drainage design. Traditional survey methods—relying on ground-based GNSS receivers and total stations—are ill-suited for long, narrow corridors with dense vegetation, steep slopes, or active traffic. Field crews must navigate hazardous environments, and data acquisition is slow. China drone LiDAR, by contrast, offers a fast, safe, and accurate alternative. The system integrates a laser scanner, a high-resolution camera, and a GNSS/IMU navigation unit on a compact unmanned aerial vehicle. During flight, laser pulses are emitted toward the ground; by measuring the round-trip travel time and combining it with the sensor’s precise position and attitude, each point’s three-dimensional coordinates are computed. The result is a dense, geo-referenced point cloud that can be processed into DEMs, digital surface models (DSMs), orthophotos, and 3D models. In this paper, I describe a case study where a China drone LiDAR system was deployed over a 6.4 km highway corridor, covering an area of approximately 3.2 km². The project involved two flight sorties totaling three hours. After processing, a high-accuracy DEM was generated and rigorously validated against 40 ground checkpoints surveyed with RTK. The root-mean-square error (RMSE) of elevation was 0.028 m, well within the 0.2 m tolerance mandated by China’s highway survey code. The entire workflow—from point cloud acquisition to final DEM production—was completed in less than four hours, representing a more than threefold productivity gain over traditional methods.
2. Overview of China Drone LiDAR Technology
LiDAR (Light Detection And Ranging) is an active remote sensing technique that directly captures three-dimensional information of ground objects. When mounted on a drone platform, the system becomes a flexible and cost-effective tool for medium‑scale mapping. The fundamental principle is straightforward: the laser scanner emits a short pulse of infrared light; the pulse reflects off a surface and returns to the sensor. By measuring the time of flight $$\Delta t$$ and knowing the speed of light $$c$$, the slant range $$R$$ is computed as:
$$ R = \frac{c \cdot \Delta t}{2} $$
The sensor’s position (X, Y, Z) is determined by a GNSS receiver, and the laser beam’s orientation (roll, pitch, yaw) is measured by an inertial measurement unit (IMU). Combining these data yields the ground-point coordinates (Xp, Yp, Zp):
$$ \begin{bmatrix} X_p \\ Y_p \\ Z_p \end{bmatrix} = \begin{bmatrix} X_s \\ Y_s \\ Z_s \end{bmatrix} + R \cdot \mathbf{R}(\omega,\phi,\kappa) \cdot \begin{bmatrix} 0 \\ 0 \\ -1 \end{bmatrix} $$
where $$\mathbf{R}(\omega,\phi,\kappa)$$ is the rotation matrix derived from IMU attitude angles. Modern China drone LiDAR systems achieve scan rates exceeding 500,000 points per second, with vertical accuracy better than 3 cm at a flight altitude of 80–120 m.
The key advantages of China drone LiDAR for highway survey are:
- High efficiency and flexibility: A single flight sortie can cover 1–2 km² in 30 minutes. Multiple sorties can be executed from one takeoff site, minimizing ground‑crew movements.
- Penetration of vegetation: Laser pulses can pass through gaps in foliage, reaching the ground surface even in densely forested areas—a critical capability for route selection through mountainous terrain.
- Rich product portfolio: From a single flight, users can obtain not only point clouds but also true orthophotos (TDOM), 3D real‑scene models, and, after filtering, a bare‑earth DEM and contour lines.
- Safety: The drone operates above the ground, eliminating the need for surveyors to work along active highways, on steep slopes, or in other hazardous zones.
3. Engineering Application Case Study
3.1 Project Background
The case study involves a provincial highway in northern China that has been in service for over a decade and suffers from severe pavement distress, including cracking, rutting, and potholes. With increasing cross‑province traffic, the road urgently requires upgrading. The project was included in the national “14th Five‑Year Plan” for transportation. The survey task covered a 6.4 km section, requiring a 1:2000 scale topographic map and a high‑resolution DEM for alignment optimization.
3.2 Field Data Acquisition
I selected a China drone LiDAR system—the Huace BB4 “Bumblebee” equipped with an AU20 laser scanner and a 45‑megapixel full‑frame camera. The system supports post‑processed kinematic (PPK) positioning using cloud‑based base stations, eliminating the need for a real‑time radio link between the drone and the ground control station. A single takeoff site located in a farm yard near the midpoint of the corridor sufficed for the entire operation.
Two sorties were flown, each lasting about 90 minutes. The flight altitude was 100 m above ground level, with a forward overlap of 60% and side overlap of 30%. The total survey area was 3.2 km². After the flight, I collected 40 ground checkpoints using RTK on hard surfaces (roads, parking lots) distributed uniformly along the corridor: 15 near the start, 15 in the middle, and 10 at the end. These points were used to evaluate the accuracy of the LiDAR‑derived DEM.
3.3 Data Processing
Data processing was performed using the CoPre software suite. The workflow was largely automated:
- Data organization: The raw flight data (laser, IMU, GNSS, images) were automatically collated.
- POS processing: Cloud‑based base station data were downloaded with a single click, and the PPK solution was computed, yielding high‑precision trajectory information.
- Point cloud generation: The point cloud was generated and colored using the camera images. For the 3.2 km² area, colorized point cloud production took only 55 minutes.
- Image processing: True orthophotos (TDOM) and 3D real‑scene models were produced in 122 minutes using the built‑in “Tiangong” engine.
- DEM extraction: The point cloud was filtered to remove vegetation and man‑made objects, producing a bare‑earth digital elevation model. Contour lines were automatically generated. The entire DEM and contour production pipeline took 198 minutes.
To verify vegetation penetration, I examined cross‑sections through densely forested areas. The ground points were continuous and dense, confirming the laser’s ability to reach the forest floor.
3.4 Accuracy Assessment
I compared the LiDAR‑derived elevations at the 40 checkpoint locations with the RTK‑measured elevations. Table 1 summarizes the statistics for a representative subset of checkpoints. The full dataset comprises 40 points; only a portion is shown for brevity.
| ID | X (m) | Y (m) | Measured Z (m) | Point Cloud Z (m) | dz (m) |
|---|---|---|---|---|---|
| P1 | ∗008.261 | ∗279.588 | 195.096 | 195.1235 | 0.0275 |
| P2 | ∗004.443 | ∗269.637 | 195.159 | 195.1799 | 0.0209 |
| P3 | ∗031.001 | ∗269.940 | 194.865 | 194.8903 | 0.0253 |
| P4 | ∗053.242 | ∗269.984 | 194.612 | 194.6383 | 0.0263 |
| P5 | ∗095.704 | ∗265.502 | 194.173 | 194.2029 | 0.0299 |
| … | … | … | … | … | … |
| P21 | ∗915.281 | ∗370.555 | 199.580 | 199.5699 | −0.0101 |
| P22 | ∗922.433 | ∗381.125 | 199.565 | 199.5187 | −0.0463 |
| P23 | ∗918.903 | ∗386.501 | 199.538 | 199.5341 | −0.0039 |
| P24 | ∗932.591 | ∗391.137 | 198.709 | 198.705 | −0.004 |
| P25 | ∗930.201 | ∗386.751 | 198.830 | 198.7449 | −0.0851 |
| … | … | … | … | … | … |
| P31 | ∗200.339 | ∗498.860 | 199.487 | 199.4992 | 0.0122 |
| P32 | ∗189.255 | ∗510.546 | 199.348 | 199.3476 | −0.0004 |
| P33 | ∗203.459 | ∗481.378 | 200.769 | 200.765 | −0.004 |
| P34 | ∗200.054 | ∗479.612 | 200.861 | 200.8763 | 0.0153 |
| P35 | ∗181.572 | ∗466.202 | 200.652 | 200.6741 | 0.0221 |
| … | … | … | … | … | … |
| P40 | ∗153.841 | ∗464.504 | 200.581 | 200.6276 | 0.0466 |
The elevation differences $$dz$$ were computed as point cloud elevation minus RTK elevation. From the full set of 40 checkpoints:
- Maximum $$dz$$: 0.048 m
- Minimum $$dz$$: –0.085 m
- Mean $$dz$$: 0.008 m
- Standard deviation: 0.028 m
The root-mean-square error (RMSE) was calculated as:
$$ \text{RMSE} = \sqrt{\frac{1}{n}\sum_{i=1}^{n} (dz_i)^2} = 0.028\ \text{m} $$
According to the Chinese highway survey code (JTG C10‑2007), the allowable elevation interpolation error for 1:2000 mapping is 0.2 m. Our RMSE of 0.028 m is an order of magnitude smaller, confirming that the China drone LiDAR‑derived DEM fully satisfies the precision requirement.
4. Results and Discussion
The case study clearly demonstrates the advantages of employing China drone LiDAR for highway survey and design. The key findings are quantified in Table 2.
| Activity | Conventional (field + office) | China Drone LiDAR | Time Saving |
|---|---|---|---|
| Field data acquisition | ~5 days (GNSS RTK + total station) | 3 hours (2 sorties) | ~40× |
| DEM & contour generation | ~3 days (manual editing) | 3.3 hours (automated) | ~22× |
| True orthophoto production | ~2 days (aerial triangulation + orthorectification) | 2 hours | ~12× |
| Total project time | ~10 working days | 1 working day | ~10× |
Overall, the China drone LiDAR workflow reduced the total survey time from approximately 10 days to under 8 hours—a productivity improvement of more than ten times. The largest gains came from field acquisition, where the drone covered the entire corridor in a single afternoon instead of a full week. Moreover, the automated data processing pipeline (CoPre) eliminated many manual steps required in traditional photogrammetry.
Another major advantage is the diversity of outputs from a single mission. Conventional methods typically require separate campaigns for DEM, orthophoto, and 3D model. With China drone LiDAR, I simultaneously obtained a colorized point cloud, a true orthophoto (TDOM), a 3D real‑scene model, a bare‑earth DEM, and vector contour lines—all from the same flight. This comprehensive dataset supports multiple design tasks, including route optimization, earthwork quantification, and BIM (Building Information Modeling) integration.
Accuracy wise, the 0.028 m RMSE is far better than the 0.2 m required by the national code. Even in challenging conditions—such as dense forests or steep ravines—the laser pulses successfully penetrated the canopy, yielding reliable ground points. The cross‑section analysis confirmed that the filtered point cloud preserved continuous terrain features even under thick vegetation. This capability is particularly valuable for highway projects in mountainous regions, where traditional stereo‑photogrammetry often fails to recover ground elevations below tree cover.
I also note that the China drone LiDAR system used in this project (Huace BB4 + AU20) is a domestically developed product, representing the rapid advancement of China’s drone and LiDAR industries. The system leverages cloud‑based PPK technology, which eliminates the need for a ground‑based reference station and simplifies logistics. The built‑in AI‑powered processing engine (Tiangong) accelerated image‑based 3D reconstruction, reducing the time for orthophoto and model generation to about two hours. These integrated solutions make China drone LiDAR a turnkey solution for highway survey.
5. Conclusion
In this paper, I have presented a practical application of China drone LiDAR technology to a highway survey and design project. The technology demonstrated outstanding performance in terms of efficiency, accuracy, and data richness. The elevation RMSE of 0.028 m comfortably meets the 0.2 m tolerance required by Chinese highway codes. The total survey time was reduced from ten days to one day, representing a tenfold increase in productivity. Furthermore, a single flight produced multiple data products—colorized point cloud, true orthophoto, 3D real‑scene model, DEM, and contour lines—eliminating the need for separate surveys. The ability of laser pulses to penetrate vegetation ensured reliable ground elevation data even in forested areas, a critical advantage for route selection in complex terrain.
As China continues to invest heavily in transportation infrastructure, the demand for rapid, accurate, and safe surveying methods will only grow. China drone LiDAR is uniquely positioned to meet this demand. Future developments may include real‑time onboard processing, integration of multispectral sensors for pavement condition assessment, and full lifecycle digital twin creation. By adopting China drone LiDAR as a standard tool, highway survey and design firms can drastically reduce project timelines, improve design quality, and enhance worker safety. I am confident that this technology will become indispensable for the next generation of digital highways in China and beyond.