As an essential foundational dataset in highway survey and design, the Digital Elevation Model (DEM) has traditionally been measured using methods such as GNSS-RTK and total stations. Highway survey and design projects often involve long distances and traverse complex terrains, making traditional DEM measurement both labor-intensive and time-consuming, with high economic costs. In our study, we applied drone technology—specifically, UAV LiDAR—to a practical highway survey and design project. By constructing a high-precision DEM from laser point clouds, we analyzed the accuracy and verified that the approach not only meets regulatory requirements but also significantly improves work efficiency.
The rapid development of the highway industry has brought increasingly severe challenges to survey and design enterprises. How to quickly formulate and propose optimal design solutions within a short design cycle has become an urgent industry problem. DEM plays a vital role in highway design projects; as an indispensable basic surveying dataset, it can faithfully reproduce the geographic environment of the project and provide accurate surveying data for planning and design. In the early stages of highway survey and design, surveying is indispensable. It plays a core role in capturing the geographic features around the project and obtaining basic surveying data. However, traditional surveying methods, due to their long operation cycle, low efficiency, and high cost, can no longer meet the increasing engineering demands with tight schedules. Especially in band-shaped long-distance routes, undulating terrain, significant elevation differences, and dense vegetation such as expressways in complex terrain conditions, traditional methods exhibit low measurement efficiency, high operational difficulty, and strict condition limitations, failing to satisfy the modern highway construction requirements for both progress and quality.
In recent years, airborne LiDAR technology, with its remarkable advantages of high precision, lightweight design, intelligence, and strong environmental adaptability, has gradually gained recognition from more and more users and has become a preferred technology in many fields, thus also becoming a research hotspot in highway survey and design. Many related studies have initially verified its technical feasibility and application potential. For instance, some researchers have applied UAV LiDAR to engineering surveys in the initial stage of highway construction and analyzed its accuracy in 1:2000 large-scale mapping. Others have elaborated on the technical characteristics of airborne LiDAR and key issues in system usage, summarizing the application of LiDAR technology in highway survey and design. Still others have pointed out that airborne LiDAR technology, with its ability to directly and quickly obtain high-precision three-dimensional information of the ground surface and strong environmental adaptability, is increasingly used in highway surveys. Studies have shown that UAV LiDAR can exert great advantages in mountainous environments with complex terrain, fragmented landforms, and dense vegetation. The laser points that reach the ground through vegetation can well represent landform features. Other research has also demonstrated that LiDAR technology has advantages such as short measurement cycles, high precision, low weather impact, and strong safety. It has been mentioned that point cloud data collected by airborne LiDAR can also be used to produce true 3D models. Using a large amount of RTK data and leveling data, the quality of LiDAR point clouds in highway projects has been evaluated, confirming that across the entire route, airborne laser data accuracy can meet the requirements for preliminary and final surveys of highway projects. Key technical points for large-scale application of airborne laser point cloud data in highway survey and design, which differ from general DEM production, have been summarized.
With the basic completion of China’s comprehensive transportation infrastructure, ensuring road network operation safety, guaranteeing smooth vehicle operation, and continuously improving service quality have become top priorities in the transportation industry. Against this background, expanding and upgrading existing highways has gradually become a key industry development direction. Highway reconstruction and expansion projects require much higher measurement accuracy than new road construction. Most existing road reconstruction and expansion surveys use low-altitude aerial photogrammetry supplemented by manual on-site measurements, which is time-consuming and labor-intensive, not only disrupting normal traffic but also posing significant safety hazards. In terms of efficiency, safety, and economic cost, it is difficult to meet the increasing demand for highway reconstruction and expansion projects in China. In our study, based on a specific highway design project, we used UAV LiDAR to obtain point cloud data within the project area and produced the required DEM results. Combined with field-measured checkpoints, we verified the accuracy of the results.
Overview of UAV LiDAR Technology
LiDAR (Light Detection And Ranging) is a remote sensing technology that has developed very rapidly in recent years. It integrates advanced technologies such as laser rangefinders and GPS. The LiDAR system actively emits laser pulses to obtain three-dimensional coordinate information of target objects. With the continuous improvement of scanner measurement accuracy and sampling rate, this technology has been widely used in fields such as forestry, ecology, botany, agriculture, and architecture. Depending on the platform, LiDAR systems can be divided into three categories: space-based, airborne, and ground-based LiDAR.
Technical Principle
The UAV LiDAR system is an active remote sensing system that integrates optics, mechanics, and electronics. Its core working principle is as follows: the system is mounted on a UAV platform. During flight, the laser scanner emits laser pulses toward the ground. By calculating the time from emission to when the laser pulse is reflected back by the surface object and returns to the platform, combined with the spatial position and attitude of the laser transmitter (obtained by joint calculation of GNSS and IMU), the three-dimensional spatial coordinates of each laser footprint can be accurately calculated. This ultimately forms a massive point cloud dataset containing geospatial information.
Technical Advantages
Compared with traditional surveying technologies, UAV LiDAR technology has the following significant advantages in highway survey and design:
- High Efficiency and Flexibility: UAVs are highly maneuverable, can be deployed quickly, and have short operation cycles. They are especially suitable for long-distance, band-shaped survey tasks, greatly shortening the field work period.
- High Precision and Density: It can directly obtain three-dimensional coordinates of ground points with centimeter-level accuracy. Point cloud density can reach hundreds of points per square meter, faithfully reproducing details of terrain and features.
- Strong Vegetation Penetration: Laser pulses can partially penetrate vegetation gaps and directly obtain the ground surface model, which is crucial for route surveys in forested areas.
- Rich Data Products: Based on point cloud data, high-precision DEM, DSM, DOM, etc., can be quickly generated to meet multi-dimensional design needs.
- High Safety: It can replace manual entry into dangerous or inaccessible areas (such as cliffs, swamps, existing operating highways) for surveys, ensuring personnel safety.
Engineering Application Case
To verify whether the DEM produced by UAV LiDAR technology can be applied to highway survey and design, we selected a highway design project section approximately 6.4 km long as the experimental area. The UAV used was a Huace BB4 Bumblebee equipped with an AU20 LiDAR and a high-definition full-frame industrial camera with a resolution of 45 million pixels. This laser aerial survey system integrates airborne LiDAR technology and photogrammetry technology, enabling simultaneous acquisition of point cloud and image data. It can not only obtain precise coordinate information of ground points through LiDAR point clouds, including ground points under vegetation, but also obtain clear image data through the high-definition camera to achieve 3D twin reconstruction of surface features. One flight can simultaneously acquire high-precision color point clouds and orthophoto (DOM) results.
Project Overview
A provincial highway has been in operation for more than ten years. The old road has many diseases, severe pavement damage, and many areas with rutting, cracking, and other defects. With economic and social development and increasing inter-provincial exchanges, the traffic demand on this section will increase year by year. The existing old road can hardly meet the growing traffic development demand, which has adversely affected regional economic development to a certain extent. To ensure the efficiency of the arterial highway and fully leverage its supporting role in economic and social development, it is urgent to upgrade this section. The upgrade of this section has been included in the key construction projects of the national and provincial “14th Five-Year Plan” for transportation, providing an excellent policy environment for this project.
Field Data Acquisition
1. Selection of UAV Takeoff and Landing Site
The AU20 base-station-free cloud service combined with post-PPK processing technology does not require continuous communication between the UAV and the ground station. Only one takeoff and landing point is needed to complete the entire survey flight mission, saving a lot of time searching for takeoff and landing sites and transferring. Based on field reconnaissance, we selected a courtyard of a village production workshop in the middle of the survey area as the UAV takeoff site.
2. Data Acquisition
After on-site equipment installation, flight parameter settings, and equipment calibration, we started the operation with one key. A total of two flights were performed, taking about 3 hours, acquiring point cloud and image data covering approximately 3.2 km² of the experimental area. After the flight, we used GNSS RTK technology to collect coordinate data of 40 checkpoints distributed at the front, middle, and rear of the project area on hard ground, for later DEM accuracy verification.
Indoor Data Processing
1. Automated Data Organization
After data collection, we used the CoPre software from Huace Navigation to automatically organize and uniformly copy the data.
2. Guided Task Management
POS solution and data processing were handled using the CoPre guided task manager, which is simple and easy to use.
3. One-Click Automatic Download of Cloud Base Station Data
CoPre can automatically download cloud base station data with one click. Compared with real-time RTK, the cloud base station is not affected by CORS coverage, network signals, or UAV communication connections, ensuring POS data accuracy and stability.
4. Fast Processing of Point Cloud and Image Data
It took about 55 minutes to complete the color point cloud data processing for the 3.2 km² survey area. CoPre includes the Tiangong engine algorithm, which enables efficient 3D model reconstruction, greatly reducing image processing time. The production of true orthophoto (TDOM) and real 3D model results for 3.2 km² took only 122 minutes.
5. Vegetation Penetration Effect Inspection
To check the point cloud penetration effect through vegetation, we examined a cross-section in a densely vegetated area. The results showed that ground points were continuous and dense, indicating good penetration.
Accuracy Analysis
To verify the accuracy of the point cloud and model results, we collected 40 checkpoints on site and compared them with the results. The accuracy verification is as follows:
For the verification of the measured laser point cloud accuracy, we measured 40 field checkpoints using RTK at the start, middle, and end of the survey area. In the office, we used LiDAR360 software to check the accuracy.
Table 1 below presents a sample of the elevation accuracy analysis for the laser point cloud (only a portion of the checkpoints are shown).
| Point ID | X (m) | Y (m) | Measured Z (m) | Calculated Z (m) | Difference dz (m) |
|---|---|---|---|---|---|
| P1 | 5008.261 | 4279.588 | 195.096 | 195.1235 | 0.0275 |
| P2 | 5004.443 | 4269.637 | 195.159 | 195.1799 | 0.0209 |
| P3 | 5031.001 | 4269.940 | 194.865 | 194.8903 | 0.0253 |
| P4 | 5053.242 | 4269.984 | 194.612 | 194.6383 | 0.0263 |
| P5 | 5095.704 | 4265.502 | 194.173 | 194.2029 | 0.0299 |
| … | … | … | … | … | … |
| P21 | 5915.281 | 4370.555 | 199.580 | 199.5699 | -0.0101 |
| P22 | 5922.433 | 4381.125 | 199.565 | 199.5187 | -0.0463 |
| P23 | 5918.903 | 4386.501 | 199.538 | 199.5341 | -0.0039 |
| P24 | 5932.591 | 4391.137 | 198.709 | 198.7050 | -0.0040 |
| P25 | 5930.201 | 4386.751 | 198.830 | 198.7449 | -0.0851 |
| … | … | … | … | … | … |
| P31 | 5200.339 | 4498.860 | 199.487 | 199.4992 | 0.0122 |
| P32 | 5189.255 | 4510.546 | 199.348 | 199.3476 | -0.0004 |
| P33 | 5203.459 | 4481.378 | 200.769 | 200.7650 | -0.0040 |
| P34 | 5200.054 | 4479.612 | 200.861 | 200.8763 | 0.0153 |
| P35 | 5181.572 | 4466.202 | 200.652 | 200.6741 | 0.0221 |
| P40 | 5153.841 | 4464.504 | 200.581 | 200.6276 | 0.0466 |
Calculating based on the 40 checkpoint elevation data, we obtained: maximum dz = 0.048 m, minimum dz = -0.085 m, average dz = 0.008 m, and root mean square error (RMSE) = 0.028 m. The RMSE was computed using the formula:
$$ RMSE = \sqrt{\frac{1}{n} \sum_{i=1}^{n} (z_{obs,i} – z_{ref,i})^2} $$
where \( z_{obs,i} \) is the calculated elevation from the point cloud, \( z_{ref,i} \) is the measured RTK elevation, and n = 40. The resulting RMSE of 0.028 m fully meets the requirement stipulated in the “Highway Survey Code” (JTG C10—2007), which mandates that the elevation interpolation RMSE should not exceed 0.2 m. Therefore, the DEM produced from UAV LiDAR point clouds is fully suitable for highway survey and design.
Results and Discussion
Through the summary of the engineering application case, we can see that the application of drone technology, specifically UAV LiDAR, in highway survey and design has very obvious advantages. These are mainly reflected in the following two aspects:
1. Highly Automated Data Processing, More Than 3 Times Efficiency Improvement
The preprocessing of data in this project was carried out using CoPre software. It took only 177 minutes to produce color point clouds, TDOM, and real 3D model results for the 3.2 km² survey area, achieving an efficiency improvement of more than 3 times compared with traditional aerial survey solutions. Then, using CoProcess software for point cloud denoising and filtering, we efficiently constructed high-precision DEM and generated contour lines with just one click. The entire workflow, from raw data to DEM and contour results, took only 198 minutes for 3.2 km². This effectively solves the problem that traditional aerial surveys cannot automatically and efficiently produce DEM and contour lines.
2. Single Mission, Multiple Outputs
In this project, a single mission obtained multiple surveying results including color point clouds, TDOM, real 3D models, high-precision DEM, and contour lines, meeting the mapping needs for topographic surveys. There was no need to use other surveying methods to invest additional manpower and time to supplement measurements. This not only greatly improves both field and office efficiency, shortening the project schedule, but also ensures higher data accuracy and quality, more comprehensive data result types, and more far-reaching application value.
Table 2 below summarizes the comparison between traditional methods and the proposed drone technology approach in terms of efficiency and accuracy.
| Metric | Traditional GNSS-RTK/Total Station | UAV LiDAR (This Study) |
|---|---|---|
| Field work time (3.2 km²) | Several days (approx. 10–15 person-days) | 3 hours (2 flights) |
| Data processing time | Manual editing, ~1 week | ~3.3 hours (automated) |
| Elevation RMSE (m) | Typically 0.05–0.10 | 0.028 |
| Vegetation penetration capability | None | Partial, ground points obtained |
| Output products | Limited (points, lines) | Point cloud, DEM, DOM, 3D model, contours |
| Safety risk | High (exposure to traffic, steep terrain) | Low (remote operation) |
Furthermore, the accuracy of the point cloud-based DEM can also be evaluated using the standard deviation and mean error. The following statistical summary provides a more comprehensive assessment.
| Statistic | Value (m) |
|---|---|
| Maximum positive error | 0.048 |
| Maximum negative error | -0.085 |
| Mean error | 0.008 |
| Root Mean Square Error | 0.028 |
| Standard deviation of errors | 0.027 |
Given that the allowable elevation interpolation RMSE for highway survey is 0.2 m, our achieved RMSE of 0.028 m is well within tolerance. This confirms that drone technology delivers superior accuracy while significantly reducing project duration and costs.
Conclusion
UAV LiDAR, as a core component of modern drone technology, has demonstrated remarkable application value in the field of highway survey and design. Its ability to efficiently and accurately acquire three-dimensional surface data provides a reliable data foundation for constructing DEM. This not only enhances data collection efficiency but also exhibits excellent accuracy, thus offering broad prospects for application and promotion. In addition, the high-precision point cloud data collected by LiDAR can support refined modeling of terrain and features in BIM three-dimensional design, effectively enabling collaborative design and visual analysis of highway projects, thereby promoting the digital and information transformation of the entire highway survey and design process. The in-depth application of this technology will help build a solid data foundation for digital highways and smart transportation, providing strong technical support for the life-cycle management of transportation infrastructure.
In conclusion, our practical experience confirms that drone technology, particularly UAV LiDAR, is a game-changer for highway survey and design. It meets the stringent accuracy standards required by codes, dramatically reduces field and office workloads, and delivers rich, multi-purpose geospatial products. As drone technology continues to evolve, its role in infrastructure development will undoubtedly expand, bringing even greater efficiency and intelligence to the transportation sector.