Effective for Public Infrastructure Management! Latest Trends in UAV Photo Point Cloud Generation

In recent years, the use of photogrammetry with drones (UAVs) to generate 3D point cloud data has been rapidly expanding in the field of public infrastructure management. UAV photo point cloud generation is attracting attention as a solution to various challenges such as maintenance of aging bridges and roads, and improving inspection efficiency and safety amid labor shortages. The technology that can generate detailed 3D models from aerial photos enables precise site records and analyses that were difficult to obtain with conventional methods, and it plays a significant role in infrastructure maintenance inspections and disaster response. The government is also promoting the use of 3D technologies, so this is truly a key technology for infrastructure management DX. This article explains what UAV photo point cloud generation is, its benefits and latest trends, and introduces points for its use in public infrastructure management. At the end of the article, we also touch on a new solution for simple surveying called LRTK, offering hints for on-site DX.

Table of Contents

• What is UAV photo point cloud generation?

• Benefits for public infrastructure management

• Latest trends in UAV photo point cloud generation

• Simple surveying with LRTK

• FAQ

What is UAV photo point cloud generation?

UAV photo point cloud generation is a technique in which a camera mounted on an unmanned aerial vehicle (drone) captures many photos from above, and software analyzes those photos to generate 3D point cloud data of objects and terrain. This method is generally called "photogrammetry" or "photo­grammetry"; software matches common feature points appearing in multiple photos and computes the 3D coordinates of each point using the principles of triangulation. The resulting point cloud reproduces the shapes of structures and terrain with high accuracy as a collection of millions to hundreds of millions of tiny points. Because photo-derived point clouds also include color (RGB) information at each point, they produce realistic 3D models that appear almost like viewing the real object.

The main advantages of photogrammetric point cloud generation are its ease of use and cost efficiency. Even without expensive laser scanners, high-resolution 3D data can be created using a handheld camera and a PC (or dedicated cloud service). With drones, wide areas can be photographed in a short time, allowing safe data capture of mountainous areas and high-altitude structures that are difficult to access from the ground. Once the captured images are loaded into specialized software, point cloud generation proceeds automatically, so it can be handled without specialist surveying skills, which has helped its spread. Thanks to recent advances in software and improvements in hardware performance, even large numbers of photos can be processed quickly on the cloud or a high-performance PC, making it possible to obtain point clouds with millimeter- to several-centimeter precision (0.04 in to half-inch level).

Note that drone photogrammetry and LiDAR (laser) surveying complement each other. UAV laser scanners are effective in cases where photos alone have difficulty capturing point clouds, such as terrain under dense tree cover or the backside of structures, while photogrammetry is cost-effective for obtaining high-density point clouds with color information. The current trend in drone surveying is to use photogrammetry and laser surveying appropriately according to site conditions to balance efficiency and accuracy.

Benefits for public infrastructure management

The 3D data obtained from UAV photo point cloud generation brings various benefits to the maintenance and management of public infrastructure. The main points are listed below.

• Improved safety: Aerial inspections by drone can inspect areas that are difficult or dangerous for people to access, such as the tops of bridges or the faces of dams, reducing the need for personnel to enter hazardous locations. Surveys at height or on steep slopes can also be conducted with reduced risk.

• Increased efficiency: Wide-area terrain surveys and condition assessments of structures can be done in a short time. For example, road or levee surveys that traditionally took several days to weeks can be completed in hours to days with drones. In one reported case, work that previously took 45 days to survey about 0.3 square kilometers was completed in just 1.5 days using UAVs. This enables both shorter construction schedules and reduced staffing.

• High-precision data acquisition: Point cloud data can record objects with millimeter- to several-centimeter precision (0.04 in to half-inch level), allowing fine deformations and dimensions to be measured accurately in digital form. If point clouds are obtained periodically, they are useful for monitoring long-term changes and deformations. Because photogrammetric point clouds include color information, surface deterioration of concrete or vegetation growth can be assessed at a glance.

• Data utilization and sharing: The 3D models obtained become intuitive tools for information sharing among stakeholders. With orthophotos (composite aerial images viewed from directly above) and point cloud models, the situation can be examined in detail from the office without visiting the site. They can be used in drawings and reports, enhancing the persuasiveness of explanations to clients and local residents. Storing the data digitally also aids future repair design and post-disaster comparisons.

• Labor savings and skills transfer: With drones and automatic processing software, sites can be recorded to a certain standard without relying on craftsmen’s tacit knowledge. Even with a shortage of expert technicians, inspections and surveys can be carried out efficiently by small teams, contributing to labor shortage countermeasures.

For example, in highway bridge inspections, drones have been used to create detailed 3D models of bridge girders and cables, allowing high-altitude inspections that used to require scaffolding to be completed quickly and safely. Also, in large-scale landslides or earthquake-affected areas, quickly generating a 3D map with drones helps in grasping the full extent of damage and planning restoration.

Latest trends in UAV photo point cloud generation

Drone photogrammetry technology advances year by year, and its applications in public infrastructure management are deepening. Here are the recent major trends.

• Higher accuracy with RTK-equipped drones: More drones are being equipped with RTK-GNSS receivers that can obtain centimeter-class position information. RTK (Real Time Kinematic) is a method that corrects satellite positioning in real time, reducing GPS errors from several meters to within a few centimeters. Because the position data of captured photos themselves become highly accurate, the accuracy of photogrammetric models improves dramatically. In many cases, acceptable accuracy can be ensured without installing a large number of ground control points, greatly reducing site preparation time and effort. Recently, RTK-capable models have become available in affordable commercial drones, enabling high-precision surveying at more sites.

• Automated analysis and anomaly detection with AI: Efforts are advancing to analyze point clouds and orthophotos generated from photos using AI to automatically detect infrastructure deterioration and anomalies. Systems that use deep learning for image recognition to detect bridge cracks, road sinkholes, or small landslide signs and alert inspectors are being developed. This enables efficient detection of anomalies that humans might overlook across vast datasets, enhancing and streamlining inspection tasks. Governments and major companies are developing infrastructure inspection systems using AI, and automation of point cloud data analysis is expected to progress further in the near future.

• Cloud services and near-real-time processing: Services have emerged that rapidly process large numbers of photos captured by drones in the cloud to generate point cloud models and orthomaps in a short time. Workflows where data are uploaded directly from the field to the cloud and processed results are checked on a tablet have become possible. This allows terrain models to be checked and measured on site without long office-based processing, supporting immediate decision-making. With improved communications environments, obtaining and sharing 3D data in near real time is a major advancement. Even without a specialist operator, a 3D model can be generated with the push of a button, dramatically improving convenience.

• Multi-sensor payloads and data fusion: Modern drones can carry not only visible-light cameras but also infrared cameras, multispectral cameras, and even small LiDAR sensors simultaneously. Combining data acquired by different sensors expands the accuracy and scope of infrastructure inspections. For example, infrared cameras can capture surface temperature distribution of a bridge to estimate degraded areas, while visible-light photos are used to create point clouds for dimensional measurements, enabling comprehensive analyses. Multi-sensor digital twinning is becoming a new trend in infrastructure maintenance. Furthermore, fusing LiDAR point clouds with photogrammetric point clouds to create colorized, high-precision 3D models enables advanced analysis through integration of multiple sensor datasets.

• Regulatory frameworks and adoption promotion: In Japan, the Ministry of Land, Infrastructure, Transport and Tourism promotes drone surveying as part of "i-Construction," and UAV photogrammetry methods are specified in public surveying manuals. Since the 2022 amendment to the Aviation Law, Level 4 operations (beyond-visual-line-of-sight flights over populated areas) have been authorized, expanding the range of drone applications in infrastructure inspections. Going forward, legal frameworks will be adjusted to enable more flexible use of drones for infrastructure management, including autonomous flights and night operations. Subsidy programs and guideline development are accelerating local government and corporate adoption, and drone photogrammetry is becoming an essential standard tool for infrastructure DX.

Simple surveying with LRTK

Alongside advanced UAV photogrammetry, recent years have seen the emergence of easy 3D surveying solutions that combine smartphones with high-precision GNSS. A representative example is LRTK. LRTK is a small device attached to a smartphone that enhances satellite positioning (RTK), giving centimeter-class position information to photos taken with the phone’s camera. By taking multiple photos with a dedicated app, a point cloud model with global coordinates can be automatically generated on the spot, and distance, area, and volume measurements can be performed immediately. This is a revolutionary tool that enables surveying work that used to require specialized equipment to be completed with just a smartphone.

Using LRTK, simple 3D measurements are possible in confined areas or indoor spaces where flying a drone is not feasible. For example, it can acquire point clouds for detailed inspections of small bridges, scans inside tunnels, or building interiors—locations where GPS does not reach or where flight is restricted—simply with a smartphone in hand. The device weighs only a few hundred grams, making it easy to carry to the site. Because the point cloud builds up on the smartphone screen in real time as you shoot, anyone can intuitively start 3D scanning, which is another attractive feature. The acquired point cloud data can be used as survey results with latitude, longitude, and elevation, eliminating the need for coordinate alignment in post-processing.

Thus, smartphone surveying with LRTK offers a new option for infrastructure management. While drones are effective for surveying large areas, smartphone surveying is powerful for additional detailed measurements and routine inspections. Without investing in expensive equipment, companies can readily start acquiring 3D data in-house, lowering the barrier to point cloud technology. By skillfully using the latest tools and incorporating the advantages of both drones and smartphones, the DX of public infrastructure management is expected to accelerate further.

FAQ

Q1. What is the difference between photo-based point clouds from drones and laser-based point clouds? A1. Photo-based point clouds are generated by analyzing images taken with a camera and are characterized by color information for each point. Laser-based point clouds (LiDAR) are obtained by directly measuring distances with a laser scanner and can capture areas that are difficult to see in photos, such as terrain under trees. Photo-based point clouds offer low-cost, high-density data but struggle where light cannot reach. LiDAR point clouds have higher equipment costs but can directly capture structural shapes with stable accuracy. It is ideal to use photogrammetry and laser surveying appropriately depending on site conditions.

Q2. What level of accuracy can drone photogrammetry achieve? A2. Under suitable conditions, drone photogrammetry can generate 3D models with accuracy within a few centimeters. By correcting the model with ground control points, accuracy in a real-world coordinate system can be ensured. Recently, cases using RTK-capable drones have increased where centimeter-level accuracy is achieved without ground control points. However, the accuracy of photogrammetry is influenced by flight altitude, camera quality, number of photos, and object shape. For high accuracy, it is important to shoot from low altitude with high resolution and ensure sufficient overlap.

Q3. Are permits or qualifications required to use drones for surveying? A3. In Japan, commercial use of drones may require flight permission or approval under the Aviation Law in some cases. For example, flights over densely populated areas or nighttime flights require prior application to the Ministry of Land, Infrastructure, Transport and Tourism. Also, the drone licensing system (skill certification) enacted in 2022 may require obtaining licenses depending on weight and flight type. While surveying qualifications themselves are not always mandatory, when submitting results as public survey deliverables you must follow procedures defined by “basic surveying organizations.” In practice, piloting skills and legal knowledge are essential for safety.

Q4. Do I always have to install ground control points for drone photogrammetry? A4. Traditionally, multiple ground control points (markers with known coordinates) have been installed to improve survey accuracy, and they are photographed for georeferencing the point cloud model. Recently, with the spread of RTK-capable drones and post-processing PPK techniques, it is increasingly possible to reduce the number of ground control points. In some conditions, required accuracy can be achieved without ground control points. However, it is important to set verification points (check points) to confirm errors in the completed model. For projects with high accuracy requirements, it is recommended to set a few ground control points even when using RTK to ensure the reliability of results.

Q5. Can 3D surveys be done with a smartphone alone, without a drone? A5. Yes. Improvements in smartphone camera performance and photogrammetry technology have made it practical to create point cloud models from photos taken with a phone. Devices like LRTK that attach to a phone to add high-precision position information while enabling real-time point cloud generation have also appeared. Smartphone-only surveying is effective for small structures, indoor surveys, and areas where drone flight is difficult. However, drones are more efficient for covering large areas at once, so it is best to choose based on the application.

Q6. What tools are needed for point cloud processing and analysis? A6. To generate point clouds from drone photos, specialized photogrammetry (SfM) software or cloud services are required. Numerous commercial and free open-source tools are available that output point clouds and orthophotos automatically when photos are uploaded. Generated point clouds can be used with dedicated viewers or imported into CAD and GIS software. Point clouds can be saved in standard LAS or XYZ formats for integration with other survey data. Recently, web platforms that allow sharing and viewing point clouds online have made it easier to utilize data internally and externally.

Q7. How will drone photogrammetry technology develop in public infrastructure management in the future? A7. Going forward, the integration of drones with AI and IoT is expected to further automate and sophisticate infrastructure inspections. For example, AI-coordinated fleets of multiple drones could perform cooperative flights to inspect bridges and tunnel interiors autonomously at night, making full automation a realistic prospect. Also, constructing digital twins (virtual replicas of infrastructure) from acquired point cloud data for continuous monitoring and degradation prediction is becoming feasible. In the future, all critical infrastructure may be 3D-modeled and regularly scanned to detect anomalies early, with AI suggesting optimal repair timings. Drone photogrammetry is expected to become a standard tool in public infrastructure management, underpinning societal safety and efficiency.