Streamlining As-Built Management with Point Cloud Scanning

Table of Contents

• Introduction: Challenges in as-built management and the need for DX

• What is point cloud scanning? (definition, acquisition methods, accuracy, etc.)

• Workflow for using point cloud data in as-built management

• Surface-based evaluation with point clouds and differences from conventional methods

• Practical use cases (pavement, structures, embankments, slopes, etc.)

• Precautions and key points for ensuring accuracy when using point clouds

• Introducing LRTK: combining easy scanning anyone can use with precision positioning

• FAQ: common field questions about point cloud scanning

Introduction: Challenges in as-built management and the need for DX

In civil engineering, “as-built management” is the construction management process of verifying and recording that completed structures and terrain conform to the design in shape and dimensions. In public works, contractors must prove quality using measured dimensional data on site against the standards specified by the client (as-built management standards). The results of as-built management are a prerequisite for inspection approval and handover, so this is a critically important task for ensuring construction quality. In long-term projects, it is also necessary to measure and record at each stage so that later-unobservable portions (such as buried items) can be proven; otherwise, their as-built conditions cannot be demonstrated after backfilling.

However, traditional as-built management has been centered on manual work and has required large amounts of labor and time. Survey technicians use tape measures, staffs, and levels to measure each critical dimension—height, width, thickness—one by one. For example, in road works, the typical procedure is to measure the width, thickness, and elevation of the roadbed at several points after completion and check whether they fall within the allowable range in the design documents. Organizing these measurement results into drawings and tables to create record books is also part of the work, placing a heavy burden on field personnel.

Additionally, manual measurement can only measure a limited number of points, making it difficult to comprehensively capture the entire as-built condition. Even when the design and the actual structure differ slightly, checks of only representative points may miss discrepancies. In practice, there are many cases where “the main measurement points passed, but another location differed from the design and was pointed out during inspection.” Busy sites also tend to suffer human errors such as forgetting to take photographs, which can lead to problems like lacking evidence for parts that were backfilled after completion.

To address these challenges, digital transformation (DX) has been increasingly expected to provide solutions. Initiatives such as the Ministry of Land, Infrastructure, Transport and Tourism’s *i-Construction* have accelerated the use of ICT technologies to improve the efficiency and sophistication of construction management. As-built management is no exception; measurement and management using the latest technology—three-dimensional point cloud data—have attracted attention. By using point cloud scanning, it becomes possible to accurately record the as-built condition of the entire site, which was difficult with conventional methods, enabling both labor savings and quality improvement. In fact, research by the ministry reported that projects that introduced ICT technologies (3D surveying, machine guidance, etc.) reduced total working hours by about 30% on average compared to conventional methods. In an industry facing labor shortages and work-style reforms, productivity improvement through DX is unavoidable, and point cloud scanning is one of the trump cards.

What is point cloud scanning? (definition, acquisition methods, accuracy, etc.)

Point cloud scanning is a measurement method that digitally records the geometry of real-world space as a set of countless points (point cloud data). Point cloud (point cloud) data contain XYZ coordinates for each point, forming a three-dimensional model as if the site were copied at full scale. While this technology has long been used in surveying and investigation, recent improvements in sensor performance and miniaturization and cost reduction of equipment have led to broader adoption on civil engineering sites.

Point cloud data acquisition methods: There are mainly two types of point cloud acquisition: laser measurement (LiDAR) and photogrammetry. In the laser method, a dedicated 3D laser scanner is used; it emits laser light and captures reflections from object surfaces to obtain high-density point clouds. For terrestrial laser scanners (TLS), a tripod-mounted unit rotates 360 degrees and can acquire hundreds of thousands of distance points per second, making it possible to obtain detailed 3D point clouds with millimeter-level precision. On the other hand, photogrammetry uses a drone or similar to take many aerial photos and reconstructs three-dimensional shapes from overlapping images using image analysis software to generate point clouds. Photogrammetry can cover large areas in a short time and is especially suitable for terrain surveys and large-scale as-built measurements. However, securing accuracy requires sufficient image overlap and the placement of ground control points (known coordinate points on the ground).

Recently, mobile mapping systems (vehicle-mounted laser scanners), handheld scanners, and even point cloud acquisition using built-in simple LiDAR on smartphones and tablets have emerged. For example, recent iPhone and iPad models include small LiDAR sensors that allow easy scanning of surrounding point clouds using dedicated apps. However, current smartphone LiDAR is inferior to professional equipment in ranging accuracy and effective range, so special measures are required to achieve the high accuracy and reliability demanded in professional work. Approaches that combine smartphones with high-precision GNSS, such as the LRTK described later, exist, but in general it is important to select appropriate equipment according to the application and accuracy requirements.

Point cloud measurement accuracy: The accuracy of point cloud scanning varies depending on the sensor, method, and environmental conditions. High-performance laser scanners can measure with errors on the order of a few millimeters, but they are large and expensive. Drone photogrammetry typically provides accuracy on the order of a few centimeters and is suitable for wide-area terrain measurement. Smartphone and simple scanners have lower accuracy—on the order of several centimeters to several tens of centimeters—but can be useful for short-range simple measurements or as auxiliary tools. In any method, accuracy can be ensured by comparing and validating against control points. It is recommended to place known control points on site and georeference the scan data to them, or to measure several check points with conventional surveying methods and compare them with the point cloud to confirm that the obtained point cloud meets the required accuracy.

Workflow for using point cloud data in as-built management

Below is a general workflow for using 3D point clouds in as-built management. The main steps are as follows.

• Planning and preparation for measurement: First plan the area and method for measuring the as-built condition. Install known control points or target markers on site as needed and define the survey coordinate system. Select measurement equipment suited to site conditions. For example, aerial drone imaging is efficient for a wide development site, while TLS is more suitable for checking the details of structures. During preparation, perform calibration and operational checks of equipment to ensure everything is ready for measurement.

• Point cloud data measurement and acquisition: Once preparations are complete, conduct point cloud measurements on site. For laser scanners, a tripod-mounted unit emits laser around the surroundings and acquires many reflection points. Change measurement positions and scan multiple times as needed to cover blind spots. Multiple point clouds obtained are later aligned (merged/registered) into a single dataset. For drone photogrammetry, capture the site from various angles and process the image set with software to generate point clouds. For large sites, flights may be divided and additional ground photos taken to cover blind spots unreachable from the air. Separate point clouds are georeferenced to a survey coordinate system using control points so that all point clouds are transformed into a unified coordinate system. This reproduces the entire site’s 3D model accurately and provides the foundational data for as-built evaluation.

• As-built analysis and comparison with design data: Use the acquired point cloud data to compare with design drawings or 3D design models and judge whether as-built dimensions are acceptable. In dedicated point cloud processing or CAD software, overlay and analyze point clouds with design data. Typical analysis methods include cross-section comparison and 3D differencing (heatmap) checks. In cross-section comparison, extract a cross-section shape from the point cloud at designated transect positions and overlay it with the design cross-section to check deviations. In heatmap analysis, compute the height difference between each point in the point cloud and the design surface automatically and produce a colored map (as-built heatmap). The color-coding makes areas higher or lower than the design immediately visible, which is effective for surface-based evaluation of excesses and deficiencies. The Ministry of Land, Infrastructure, Transport and Tourism has recently begun incorporating such surface-based measurement data like point clouds into guidelines for as-built evaluation (“surface management”). Instead of limiting evaluation to point-by-point numerical checks as in conventional methods, judging acceptability over an entire surface increases inspection coverage and reliability.

• Record creation, reporting, and delivery: The as-built information obtained from point cloud analysis is compiled into record documents and reported to the client. In some cases, submissions are made in traditional forms such as drawings and numerical tables following electronic delivery guidelines; increasingly, however, entire 3D point cloud data or 3D models reflecting the differences between point clouds and design (for example, IFC or LandXML formats) are delivered as electronic data. Point clouds can be preserved long-term as “digital evidence,” providing more reliable as-built records than photo albums or cross-section sketches. If additional works or claims arise in the future, saved point clouds can accurately reproduce the construction status at that time, eliminating the need to re-survey the site. Thus, as-built point cloud data are valuable not only for inspection records but also for future maintenance and dispute prevention, making them an asset beyond mere documentation.

The above outlines the basic flow of as-built management using point clouds. In summary, the main feature is that “the entire site is 3D-measured, and you can freely extract cross-sections or dimensions from any location afterwards.” The worry of missing a measurement due to human oversight is reduced, improving efficiency in inspection attendance and simplifying document preparation.

Surface-based evaluation with point clouds and differences from conventional methods

As mentioned above, the introduction of point cloud scanning significantly shifts as-built management from “measuring points” to “measuring surfaces.” Conventionally, experienced surveyors selected important locations and measured point-by-point, inferring and evaluating the whole from that limited data. With point cloud use, the entire site shape can be measured comprehensively at once, greatly reducing local oversights. Heatmaps that visualize differences between design and as-built are particularly useful for intuitively finding and correcting construction errors and surface irregularities.

Consider pavement as a concrete example of as-built management. Traditionally, to confirm pavement thickness, coring holes were taken at several points on the surface after construction and thickness measured. Using 3D point clouds, however, the height distribution across the entire paved surface can be measured and flatness evaluated on a surface basis. This enables a stricter grasp of pavement quality than before. Some point cloud analysis software can automatically calculate differences from the design and perform pass/fail determinations, enabling semi-automation of inspection tasks. Subtle defects that human inspections had missed can be extracted from the data, dramatically raising quality control levels.

At the same time, new challenges accompany the introduction of point cloud measurement. For example, assembling high-performance equipment incurs costs, and handling massive point cloud datasets requires attention to PC performance and data management methods. However, these challenges can be overcome with the countermeasures described in the next section, and the benefits are substantial. Surface-based evaluation using point clouds has been included in draft guidelines by the ministry and is being trialed not only in earthworks and pavement but also in bridges, tunnels, dams, and other work types. As a new norm in as-built management, the practice of “measuring surfaces” is expected to spread further.

Practical use cases (pavement, structures, embankments, slopes, etc.)

Point cloud scanning for as-built management is being put into practical use across many work types and scenes. Below are representative use cases.

• Pavement examples: As described above, in road paving works, full-surface checks of finished surface flatness using point clouds are being conducted. Where sampling inspections like coring for pavement thickness were common, point clouds allow measurement of the entire pavement surface immediately after construction and evaluation of deviations from standards via heatmaps. In one site, introducing point clouds reportedly reduced the days required for as-built inspection to less than half of previous times. Automatically generated color maps from point cloud data indicated unevenness, allowing instantaneous identification and correction of areas requiring rework. Clear data-based proof of pavement quality provided reassurance to both clients and contractors.

• Structure examples: Point cloud as-built management is expanding in structural fields such as bridge substructures, tunnels, and dams. For example, in bridge foundation work, the positions of driven piles and dimensions of exposed rebar can be recorded with point clouds and checked against 3D design data (BIM models). This enables detection of slight distortions or misalignments in concrete structures that are difficult to find with manual measurement, leading to timely corrections and quality assurance. Initiatives also include surface-based checks of tunnel lining concrete thickness using point clouds to confirm uniformity. For large structures, manual measurement has clear limits, but point clouds enable detailed as-built understanding in short time, prompting major contractors to adopt the technology actively.

• Embankment and earthwork examples: In earthworks (embankment and excavation), point clouds are actively used for both as-built management and volume control (calculation of embankment/excavation volumes). A regional mid-sized civil engineering firm used drone photogrammetry services to record the as-built of an embankment as point cloud data, greatly streamlining surveying work compared to conventional methods. While manually surveying wide areas is time-consuming, drones complete it quickly and volumes can be automatically computed from point clouds and reflected in as-built tables. Some sites display excess or deficit earth volumes against the as-built reference surface as a heatmap, making overfilling or overexcavation immediately apparent. These practices reduce rework and improve inspection efficiency, contributing to overall project cost management.

• Slope examples: Point cloud scanning is also powerful for steep slope (retaining slope) works. Traditionally, measuring slope as-built required survey staff to climb dangerous slopes and measure points individually, posing safety and workload issues. With point clouds, the entire slope can be measured remotely and without contact, enabling safe and comprehensive as-built management. In practice, terrestrial laser scanners placed at the bottom of a slope can acquire point clouds in minutes, and the design surface deviation of every point can be checked to determine whether it falls within standards (for example, within a tolerance of ±5 cm (±2.0 in)). In one prefecture, all local civil engineering offices introduced 3D point cloud processing systems and began actively applying point cloud inspections even to small slope works. Because it reduces dangerous work while increasing inspection accuracy, municipalities as clients have recognized its effectiveness and acceptance of point cloud data submission has spread.

Precautions and key points for ensuring accuracy when using point clouds

Although point cloud scanning is convenient, there are issues to be aware of when introducing and operating it. Here are the main precautions and countermeasures.

• Ensuring measurement accuracy: Point cloud measurement accuracy is affected by the characteristics of the equipment used and environmental conditions. To obtain high-accuracy results, pre-measurement equipment calibration and scanning at appropriate resolution and measurement distances are important. As mentioned earlier, compare post-acquisition data with known control points to check for errors or distortions in the point cloud. For example, in drone photogrammetry, placing a sufficient number of ground targets to correct GNSS errors is effective; in TLS measurements, include multiple known coordinates within the range. The Geospatial Information Authority of Japan also recommends cross-validation with conventional methods when introducing new technologies, so making cross-checks a habit on site is crucial.

• Handling large-volume data: Point cloud datasets often contain very many points, leading to huge file sizes. One site can reach hundreds of millions of points (several GB), imposing burdens on storage, sharing, and PC processing. Countermeasures include thinning unnecessary points to reduce data volume, generating polygon meshes or contour lines from point clouds to lighten the data, and using cloud services that display and edit point clouds so large datasets can be handled via a browser without a high-end workstation. For long-term storage, pay attention to future format compatibility. Saving in industry-standard formats such as LAS or E57 will make it easier to read even if software changes.

• Equipment and software introduction costs: Fully introducing point cloud technology requires significant initial investment in 3D laser scanners, high-performance drones, and analysis software. Typical laser scanners cost several million to tens of millions of yen, and with software and maintenance contracts the initial cost can approach 10 million yen in some cases. Even drone photogrammetry with high-precision aircraft and dedicated software can cost several million yen. These are major obstacles for small and medium enterprises, but equipment rental and cloud services (monthly subscriptions) are becoming available to lower the entry barrier. Some municipalities also provide ICT introduction subsidies. Start with small sites using inexpensive means and verify effectiveness before expanding adoption gradually.

• Human resource development and operational structure: Mastery of point cloud technology requires acquiring skills in equipment operation and data processing. Although modern equipment has become easier to operate, foundational knowledge and experience still affect result quality. If knowledgeable staff are not available internally, take advantage of manufacturer or specialized training courses. Assigning responsibility for new tools to younger employees with veteran support and repeatedly comparing results with conventional methods can improve organizational skills. Standardizing internal rules for handling point cloud data and deliverable creation is also important. To avoid situations where acquired data is not utilized, establish information-sharing systems between departments.

• Appropriate technology selection: Point cloud measurement encompasses various methods—laser scanners, UAV photogrammetry, mobile mapping, ground photogrammetry (camera + software), etc.—each with different accuracy characteristics and strengths. Choosing the right method for site conditions and objectives is essential. For example, in forested areas laser is often needed to obtain ground surface, while for confined indoor spaces a handheld scanner can be better than a drone. The accuracy required for as-built management (comparable to leveling measurements or at the millimeter level) should guide method selection. Verify via small-scale trial measurements that the chosen method can meet the as-built standards set by the client. In some cases, consider combining conventional methods: use manual point measurements to guarantee critical dimensions and record overall shape with point clouds to play to each approach’s strengths.

These are the main cautions when using point clouds. While challenges exist, appropriate countermeasures make point cloud technology a substantial asset on site. These are the kind of hurdles typical to new-technology introduction periods. Start small within your capacity, experience the benefits on site, and scale up gradually. The ministry’s officials also encourage starting with any technology that is easy to begin with. Take one step at a time and advance DX on your sites.

Introducing LRTK: combining easy scanning anyone can use with precision positioning

To broaden the uptake of point cloud scanning, we introduce LRTK here. LRTK is a positioning and measurement system developed by Reflexia, a venture spun off from Tokyo Institute of Technology, characterized by the fusion of “easy 3D scanning anyone can use” and “cm-level precision positioning.” Traditionally, obtaining millimeter-accuracy 3D point clouds required expensive laser scanners and expert knowledge. LRTK leverages the LiDAR sensor built into commercial smartphones (iPhone) while applying proprietary GNSS (satellite positioning) technology to attach high-precision positional information to measurement data in real time. As a result, it can obtain 3D point clouds with accuracy comparable to specialized equipment while operation is intuitive via a smartphone app, making it accessible even to less experienced technicians.

Because LRTK is a compact, lightweight smartphone-based system that is easy to carry, it can flexibly be used in confined sites and both indoor and outdoor measurements. Acquired point cloud and positioning data are automatically uploaded and managed in the cloud, enabling real-time sharing between the field and the office.

In addition to point cloud scanning, LRTK includes AR (augmented reality) functions that allow overlaying design models on the smartphone screen during construction for checks, or supporting stake placement tasks such as guiding pile-driving positions. As an integrated platform that realizes these functions in one package, LRTK is attracting attention as an alternative to traditional expensive surveying equipment and as a solution to labor shortages, with adoption already underway in municipalities and construction companies. Price information is not disclosed here, but because LRTK significantly lowers the adoption barrier compared to conventional equipment, it can be regarded as an innovative tool that democratizes point cloud scanning. Those interested should check LRTK’s official site for materials and demo videos.

FAQ: common field questions about point cloud scanning

Q1. What equipment is needed to start point cloud scanning? Is it impossible without an expensive laser scanner? A1. There are various ways to acquire point cloud data depending on the application and scale. High-performance laser scanners are desirable for large sites requiring millimeter accuracy, but drones and smartphone LiDAR can handle small-scale as-built measurements these days. You can also outsource measurement to external surveying companies and obtain only the point cloud data. Solutions that combine a smartphone with GNSS, such as LRTK, now provide an affordable path to high-precision point clouds. Start by testing with rental equipment or free apps to determine the method that suits your company.

Q2. Point cloud measurement seems difficult—are special skills or qualifications required? A2. No special license is required, but basic operational skills for equipment and software must be acquired. Modern surveying equipment has user-friendly interfaces, and most can be operated after short training. Manufacturer or municipality-run training sessions are held in many locations, and products that offer beginner-friendly simple apps—like LRTK—are available. Start by working alongside experienced technicians and gradually build proficiency. Learning basic point cloud concepts (coordinate systems, survey error concepts, etc.) in advance will deepen understanding.

Q3. Can point cloud as-built data be trusted for accuracy? Might it contradict conventional measurements? A3. When planned and executed properly, point cloud measurements are sufficiently reliable in terms of accuracy. Using high-precision equipment, correcting with control points, and cross-checking with existing surveys can yield accuracy comparable to conventional tape and level surveying. Because point clouds can cover the entire site, they excel in both accuracy and comprehensiveness. That said, sensor characteristics can cause some errors or missing data in low-reflectivity materials or occluded areas. For critical points, combining with conventional methods, performing multiple scans, or cross-validating will ensure practical accuracy and reproducibility.

Q4. Will clients (inspectors) accept point cloud data submissions? Don’t I still need to prepare photo albums and cross-section drawings as before? A4. For nationally managed projects, guidelines for as-built management using 3D data have already been established, and inspections using point cloud data are underway. Local governments are increasingly accepting point cloud submissions as part of ICT-enabled works. However, during this transitional period, many projects still require traditional drawings and records in parallel with point clouds. In practice, it is common to use point cloud analysis software to automatically output cross-section drawings and as-built pass/fail tables for submission, thereby meeting documentation requirements while using point clouds. Acceptance of 3D data by clients is expanding, and it is expected that in the future inspections will increasingly be completed by delivering point clouds or 3D models electronically.

Q5. Handling massive point cloud data is intimidating. I’m worried about PC specs and software costs… A5. Point cloud processing requires adequate PC performance, but there are ways to manage this. Initially filter or thin unnecessary points to reduce file size and load only required ranges during analysis to reduce processing load. Cloud services for viewing and editing point clouds have emerged, allowing use of high-performance servers over the internet. These services are often available on a monthly subscription, which can be cheaper than purchasing software. There are also several open-source and free point cloud viewers to try basic visualization and analysis. If procuring high-spec PCs internally is difficult, leveraging external resources makes handling large datasets feasible today.