As-built Management Changed by SfM Processing: Achieving Improved Accuracy and Efficiency on Site

1. What As-built Management Is, and Traditional Challenges (Limits of Survey Points, Labor, Work Time)

As-built management is the construction management process of verifying that completed structures and terrain conform to the shapes and dimensions in the design drawings, and of recording and reporting the results. It is key to ensuring quality and is an essential step for passing inspections and handover. Traditionally, site supervisors and surveyors measured prescribed survey points using measuring tapes, levels, total stations, etc., compared the results to the design values, recorded measured values in field notebooks, and later compiled charts and reports in the office.

However, this traditional as-built management has several identified issues. The main problems are as follows:

• Limit on number of survey points: The number of points that can be measured manually is limited, making it difficult to comprehensively capture as-built conditions over wide areas. Inspections necessarily focus on limited points, creating the risk of overlooking subtle undulations or out-of-spec areas. For example, when checking the as-built condition of a roadside slope, measuring only a few representative cross-sections cannot capture the three-dimensional finish of the entire slope, and it is common in later inspections to be told, “the main survey points were within tolerance, but there were errors in the intermediate areas.”

• Burden of labor and work time: Traditional surveying often requires a multi-person team and can take a full day. Manual measurements with tapes and levels or the setup and observation of total stations require time and effort, and the larger the site or structure the more man-hours are consumed by measurement and record organization. With increasing labor shortages, this effort and personnel requirement has become a major burden on sites.

• Accuracy, reliability, and recording errors: Manual recordkeeping carries risks of human error. If notes are incomplete, transcription errors occur later, or photos are forgotten, the reliability of measurement results can be compromised and quality issues can arise. As long as as-built management relies on measurements at isolated points, it is difficult to grasp the whole picture and errors are likely, which causes significant stress for site engineers.

As described above, traditional as-built management methods required substantial time and labor while covering only limited areas, posing challenges in both efficiency and accuracy.

2. How Photogrammetry with SfM Processing Enables Surface-based As-built Management

Recently, advances in digital technology have begun to significantly change how as-built management is done. The Ministry of Land, Infrastructure, Transport and Tourism promotes *i-Construction* and *ICT construction* and encourages the use of 3D surveying technologies such as drone photogrammetry and 3D laser scanners on sites. Among these, photogrammetry technologies that generate high-density 3D point cloud data from photos taken by drones or cameras are attracting attention. Using *SfM (Structure from Motion) processing*, a representative method, features are extracted and matched from many overlapping photos to simultaneously estimate camera positions/orientations and 3D shape, producing a point cloud model that realistically reproduces the entire site.

The point cloud data obtained by photogrammetry is a collection of countless points that construct the space—a 3D model that looks like a copy of the site. Each point contains XYZ coordinates (position) and RGB color information, recording surfaces of terrain and structures with high precision. This 3D measurement makes it possible to capture as-built conditions in a surface and volumetric manner, whereas traditionally only partial areas could be measured. In other words, because construction areas can be measured at high density down to every corner, as-built evaluation can be performed “by surface rather than by point.”

For example, in paving work, traditionally only a few points per section were checked for pavement thickness. Using point cloud data, however, the overall finished surface flatness can be evaluated with a heat map display. You can visualize as-built conformity to the design across the whole surface as color distributions, instantly seeing where material is excessively piled or lacking. The Ministry’s as-built management guidelines (revised 2022) include “surface management” methods using 3D surveying technologies, and these are expected to prevent missed inspection points and enhance quality control.

Thus, by utilizing photogrammetry with SfM processing, wide-area as-built conditions can be digitally measured in a short time, allowing accurate capture of details previously overlooked. As a result, as-built management is transforming from a “manual process of measuring scattered points over a small area” to a “process of measuring the entire site and processing the data.”

3. Operational Example: Generating Point Clouds by Drone Aerial Photography + SfM Processing, Comparing and Displaying Differences with the Design Model

Let us look at the actual workflow of drone aerial photography combined with SfM processing for as-built management. For example, at an embankment roadwork site, after construction completion a drone was used to photograph the construction area from the air, obtaining many photos. Analyzing these photos with an SfM algorithm generates a high-density 3D point cloud model representing the current state of the embankment.

Next, the 3D design model or completion design data is prepared and overlaid on the point cloud to perform as-built comparison. Using dedicated software or cloud services, overlaying the design model and the current point cloud allows overall verification of whether the finished state matches the design. With a difference display function, areas that match the design can be colored green, areas that are high beyond the tolerance can be colored red, and low areas blue, automatically color-coded. This makes it possible to spot construction defects or areas of excessive fill/cut at a glance and immediately decide where rework is needed.

Specifically, the height difference between the point cloud and the design surface is calculated and color-coded according to that difference. For example, by setting parts that are +5 cm (+2.0 in) or more high to red, parts that are -5 cm (-2.0 in) or more low to blue, and parts within the allowable range to green, pass/fail evaluation of as-built quality based on numbers becomes immediately obvious. Because this enables quantitative verification rather than relying on seasoned intuition, self-checks before inspections become reliable. Furthermore, by calculating the volume difference between design and current conditions, you can instantly determine “how many cubic meters of soil must be removed to match the design.” Such comparison of drone + SfM point clouds with design data evolves as-built management into a digital inspection that views the entire site.

4. Efficiency Methods: Automatic Cross-section Generation from Point Clouds, Heat Maps, Volume Calculation, etc.

By utilizing point cloud data obtained from photogrammetry, various as-built management tasks can be greatly streamlined. Representative utilization methods include:

• Automatic cross-section generation: Vertical and transverse sections can be cut out at arbitrary positions from the point cloud and cross-section drawings generated automatically. Cross-sections that used to be measured on site can now be obtained freely from digital data. For example, you can continuously draw cross-sections every 10 m (32.8 ft) for roads or levees, or later add checks of complex terrain cross-sections, all with a single button click. The time to create as-built drawings is dramatically shortened and comparison with design drawings is made easier.

• As-built evaluation with heat maps: As mentioned earlier, displaying differences between the point cloud and the design model as an overall heat map enables quality control via “surface visualization.” Localized errors are not overlooked and finish variability can be quantitatively evaluated. Checks like pavement flatness or slope gradients can be judged intuitively by color, increasing the persuasive power of inspection documents.

• Volume and quantity calculation: Volumes such as fill and cut can be rapidly calculated from point cloud data. By comparing pre- and post-construction terrain point clouds to compute earthwork volumes, this also aids progress quantity control. Even in cases where quantities were previously estimated from limited sections, point clouds allow accurate quantities based on the entire current state to be obtained in a short time. Calculations of thousands of cubic meters can be completed in software in seconds to minutes, accelerating quantity aggregation and progress reporting.

In addition to these, a variety of digital measurement applications are possible, such as directly measuring dimensions, slopes, and areas on the point cloud, or generating high-accuracy orthophotos (composite overhead aerial images) to use as drawing backgrounds. Using cloud services that handle point cloud processing, data collected on site can be uploaded on the spot and office-side analysis and drawing generation can begin immediately. The DX of as-built management (digital transformation) dramatically reduces recordkeeping effort and greatly smooths information sharing among stakeholders.

5. How LRTK Contributes to Eliminating or Simplifying GCPs for Correction (Unifying Coordinate References, Auxiliary Observations with Smartphone GNSS, Ground Photography in Occluded Areas)

A challenge in improving photogrammetry accuracy is giving the captured images accurate position coordinates. In conventional aerial photogrammetry, it has been common to install multiple known-coordinate markers on the ground called GCPs (Ground Control Points) and use them to correct the model coordinates. However, installing and surveying numerous GCPs takes time and effort and is a burden on the site.

One solution that has become widespread in recent years is direct georeferencing using RTK-GNSS. Among these, an innovative approach is LRTK, a technology that brings high-precision positioning to smartphones. LRTK is a system consisting of a small RTK-GNSS receiver developed by Refexia and a dedicated app, which is attached to a handheld smartphone such as an iPhone or iPad. Combining smartphone convenience with RTK positioning accuracy, it is attracting attention as a pocket-sized surveying instrument that enables centimeter-level positioning (cm level accuracy (half-inch accuracy)) anywhere by anyone.

By using LRTK on site, it is possible to unify the coordinate reference of drone photogrammetry and greatly reduce the need to install GCPs. For example, using RTK-equipped drones or drone images corrected with LRTK provides high-accuracy absolute coordinates to the point clouds obtained from aerial photos from the start. As a result, the as-built point cloud can be aligned with existing survey coordinate systems without placing many ground control markers. Time-consuming coordinate alignment work becomes unnecessary, and the speed from shooting to point cloud generation is dramatically improved.

Also, smartphones equipped with LRTK are effective as auxiliary observation tools on site. Areas that become occluded to drones—under bridge girders or under tree canopies—can be photographed or scanned from the ground with a smartphone. The point clouds and photos acquired with the smartphone include RTK position information, so they are automatically overlaid on the drone point cloud and handled as a single model. This ensures that details the drone cannot cover are reliably measured, enabling complete 3D data capture of the entire site. Parts that previously required separate terrestrial laser scanning or manual measurements can now be easily covered with an LRTK smartphone.

Furthermore, if you perform control point surveying on the smartphone using LRTK and obtain coordinates of known points, it is easy to align the point cloud model to those coordinates in post-processing. Without expensive dedicated equipment, measurement of control points and as-built measurement can be handled with just a smartphone and a small device, making it easier for technicians without specialized surveying knowledge to operate. LRTK is a powerful solution that can eliminate or minimize GCPs and simplify coordinate correction.

6. Balancing On-site Responsiveness and Speed through LRTK Drone–Smartphone Integration

The true value of LRTK lies in dramatically increasing on-site responsiveness and speed by flexibly combining drone surveying and smartphone surveying. The drone can rapidly capture wide areas from above, while the smartphone can easily capture detailed information from the ground—LRTK’s strength is being able to use these two approaches as needed and unify the data.

For example, at a large development site, start by acquiring an overall point cloud via drone aerial photography to grasp the broad as-built conditions. For detailed structures or areas not visible from above, workers can walk the site with a smartphone in hand and perform supplementary measurements so nothing is missed. Because the two data sets are integrated in a common RTK coordinate system, the complete 3D model of the site can be finished the same day. What used to take survey teams several days can, by leveraging LRTK drones and smartphones, in some cases be completed by one person in about half a day.

In addition, cloud integration enables near-real-time information sharing between the site and the office. Positioning data and point clouds uploaded from the LRTK app can be immediately viewed on cloud viewers or maps, allowing remote offices to begin as-built checks and drawing creation almost in real time. There is no need to bring data back on USB media—measure on site and share on site, enabling a speedy workflow.

Thus, LRTK-based drone and smartphone integrated surveying is a surveying style that balances accuracy and efficiency while responding quickly to changing conditions. It allows measurements to be completed in short windows between weather windows or work tasks, and if additional measurements are needed they can be handled immediately with a smartphone. Because heavy equipment and many personnel are not required, the system can respond nimbly to ad hoc measurement requests and accelerate the PDCA cycle of construction management.

7. Naturally Guiding the Introduction of Simple Surveying with LRTK as a Technology Directly Linked to Improved Site Quality and Productivity

As we have seen, the digitalization of as-built management using SfM photogrammetry and LRTK is a groundbreaking technology directly linked to improved site quality and productivity. Surface-based as-built capture with point clouds prevents missed construction errors and ensures quality, while streamlining and reducing labor in surveying achieves substantial time savings and personnel reductions. What had been limited to some large-scale projects or specialist vendors can now be easily introduced to small- and medium-sized sites by leveraging smartphones and the cloud.

Indeed, the 2022 revision of the Ministry’s standards, which officially positioned simple 3D measurement with smartphones as part of as-built management, has accelerated this trend. It is now possible to meet accuracy requirements for as-built measurement with just a smartphone and a small device, and public and private sectors alike are expanding use. This goes beyond merely adopting the latest gadgets; it is an effective countermeasure to worsening labor shortages and a support measure for young engineers, contributing to the construction industry’s overall DX promotion.

Finally, simple surveying using LRTK has made “affordable, anyone-can-use, single-person as-built management” a reality. This solution, which reduces site burden while strengthening quality control, has the potential to become the new standard. If your company is considering improving as-built management efficiency or adopting 3D technologies on your sites, we recommend positively considering the introduction of smart surveying with LRTK. It is an opportunity to harness digital technology to dramatically improve on-site productivity and quality.