CHCNAV: Expanding a Highway in China

Due to China’s rapid growth, the G85 highway, which opened in 1995 and connects Chongqing to neighboring provinces, in 2023 required expansion to four lanes. Like with any construction project, the first step was a survey. When the highway was built, surveyors had to rely on total stations and other optical instruments. Today, despite the availability of GNSS receivers, surveying over long distances in rugged terrain is still challenging.

Orthophoto of the service area in the section of the G85 highway that is being enlarged. (Photo: CHCNAV)

Li, a surveyor responsible for surveying a 5 km section that included a service area, bridges, culverts, and embankments, wanted to avoid closing lanes, which would have been expensive and dangerous due to heavy traffic. Additionally, using only GNSS receivers and total stations to complete the project would take a long time and potentially require multiple surveys. Instead, he opted to conduct a lidar survey.

To meet the project’s 2 cm root mean squared error (RMSE) accuracy requirement, Li established ground control points (GCPs) before scanning. To avoid disturbing the traffic and ensure safety, he placed the GCP targets within 50 m of the roadside. Then, a 50-minute flight was enough to scan the 5 km section.

The data was then imported into CHCNAV’s CoPre lidar processing software, which performed point cloud correction and bundle adjustment, increasing the absolute accuracy of the road surface point cloud to the required 2 cm. Next, the software performed point cloud classification, modeling, point cloud coloring, and image georeferencing and generated depth maps.

The resulting color point cloud clearly shows road markings and other features, and makes it possible to accurately measure the locations of drainage ditches, slopes, and culverts. For power lines crossing the highway, the point cloud provides accurate measurements of the minimum distance between the lines and the road for safe equipment operation.

Lidar scanning captures detailed ground surfaces, but road design relies on actual terrain conditions. Using CHCNAV’s CoProcess post-processing software — which has built-in adaptive ground point filtering algorithms — the team removed vegetation, guardrails, and vehicle returns, revealing the bare ground for design. They also accurately extracted road features, including dashed and solid lane lines with width and line type parameters, to enhance the efficiency of subsequent design efforts.

Lidar point clouds provide much richer ground detail than traditional surveys. This allows CoProcess software to automatically generate cross-sections from processed point clouds, while manual editing options are available for special terrain, such as roadside ditches. Sections can be exported to design formats or CAD drawings for immediate use.

For this project, two engineers performed the field scanning, and one engineer handled the point cloud processing, classification, and modeling to provide multi-dimensional data that met the 2 cm accuracy criteria.