Technical Explanation of RTK and Point Cloud Scanning

Positioning Principles of Laser Scanning and Photogrammetry

In point-cloud surveying using laser scanners, the device emits laser pulses toward an object and measures the time it takes for the pulses to reflect back. The distance to the object is then calculated from this measured time difference.

Three-dimensional coordinates are calculated from the laser emission angle and distance measurements, capturing the object's surface as a collection of millions of individual points. This method offers highly accurate measurements for each individual point, enabling precise recording of building shapes or terrain with millimeter- to centimeter-level accuracy in many cases. On the other hand, photogrammetry analyzes positional relationships between distinctive points (feature points) of an object that appear in multiple photographs. By identifying common features from images captured at different viewpoints, photogrammetry employs triangulation principles to determine their 3D coordinates. Since photographs also contain RGB color information, point clouds and 3D models produced through photogrammetry benefit from realistic textures, colors, and patterns matching the actual object. Photogrammetry typically offers a more cost-effective solution compared to laser scanning and is well-suited for modeling extensive terrains and large structures.

The generated models include latitude, longitude, and elevation information for each component, allowing immediate on-site measurement of distances, areas, and volumes, as well as the ability to create sectional views for drawing purposes.

​

How RTK-GNSS Positioning Correction Works

In RTK-GNSS positioning, satellite positioning data obtained from a base station (reference receiver) is compared in real-time with data from a mobile receiver (rover). By subtracting common error components between the two datasets, it achieves highly accurate positional determination.

Specifically, the base station has a known, accurately surveyed coordinate location. It transmits correction data derived from the differences between its own satellite observations (e.g., GPS, GLONASS, QZSS) and those observed by the rover. The rover receives this correction data and integrates it into its own positional solution, enabling real-time calculation of coordinates with accuracy within just a few centimeters.

In network RTK, correction data is received over the internet (e.g., via the Ntrip protocol) from sources such as the Geospatial Information Authority of Japan's electronic reference station network or private-sector GNSS base station networks. This approach allows centimeter-level positioning accuracy across extensive areas wherever network connectivity is available. Unlike Post-Processed Kinematic (PPK), which achieves similar accuracy but requires post-processing, RTK provides immediate, high-precision positioning results, offering superior real-time responsiveness.

In particular, when georeferencing point cloud data obtained from technologies such as LiDAR, it is highly recommended to aim for RTK-level accuracy wherever possible.

​

Method for Enhancing Point Cloud Accuracy Using LRTK

LRTK is a solution developed to facilitate easy utilization of the RTK-GNSS technology described above. For example, Lefixea's "LRTK Phone" is an ultra-compact RTK-GNSS receiver that seamlessly integrates with an iPhone or iPad, transforming a single smartphone into an advanced surveying device.

When capturing point clouds using a smartphone’s camera or LiDAR sensor, centimeter-level location data obtained from the LRTK device is embedded directly into the photographs or point cloud data.

Specifically, an LRTK reference station is set up within the measurement area, and its high-precision coordinates (such as latitude, longitude, and altitude in a public coordinate system) are captured as reference targets within images taken by the smartphone camera.

First, a reference device (such as an antenna with a distinct red marker) is captured in the image. Then, by scanning surrounding structures or terrain afterward, the entire point cloud dataset automatically aligns with the public coordinate system.

Traditionally, manual alignment (ground calibration) with known points or placing multiple Ground Control Points (GCPs) and correcting data shifts through post-processing was necessary for this coordinate alignment.

However, by using LRTK, position correction occurs simultaneously with point cloud acquisition, significantly reducing additional alignment work. This also means RTK data can simultaneously handle lens distortion corrections and geo-tagging during photogrammetry. The resulting point cloud data, possessing high absolute-coordinate accuracy, proves highly effective during integration with BIM or GIS, as described later.

​

On-Site Applications

Application in Construction Management and As-Built Documentation

At civil engineering and construction sites, it's essential to accurately measure completed earthworks (such as embankments and excavations) for quality control and work volume calculations. Point cloud scanning has revolutionized this as-built management process. With point cloud data captured through drone photography or terrestrial laser scanning, volumes of embankments and slopes can be instantly calculated. By combining this process with RTK positioning, the reliability of these measurement results is further enhanced.

For example, guidelines from Japan's Ministry of Land, Infrastructure, Transport and Tourism specify that "the positional accuracy of 3D point clouds used in as-built management must be within 5 centimeters." Utilizing RTK-GNSS technology makes it possible to efficiently achieve this high-precision standard.

In actual on-site applications, devices such as LRTK streamline the workflow: first setting up reference points, then performing point cloud scanning, and finally automatically generating accurate as-built point clouds. This procedure allows faster, more precise measurements compared to traditional methods. As previously illustrated with smartphone-based measurements, simply placing the RTK antenna ("RWP") on the ground and scanning embankments with an iPhone can yield point cloud data aligned seamlessly with public coordinate systems.

By precisely digitizing structures both during and after construction, differences from the original design can be easily verified, facilitating the preparation of progress reports and significantly enhancing the efficiency of on-site inspections.

Integration of Point Cloud Data with BIM

In the fields of architecture and civil engineering, the use of BIM/CIM (Building/Civil Information Modeling) is becoming increasingly widespread. BIM models represent 3D designs created during the planning stage, but verifying their consistency with actual construction results often requires overlaying measured point cloud data onto these models. If the coordinates of the point cloud data are inaccurate, aligning them with the BIM models can be time-consuming and challenging. However, with RTK-assisted point cloud scanning, the collected point cloud data itself is recorded directly within the same coordinate system as the BIM design, enabling nearly automatic alignment with the BIM model.

For example, a 3D model created using photogrammetry with LRTK has absolute coordinates derived directly from RTK-GNSS for each point, enabling automatic and precise positioning and orientation when imported into BIM models or GIS maps.

As a result, users can measure dimensions and verify differences from the design model remotely from the office, as if they were physically present at the construction site.

For instance, overlaying the scanned point cloud of a structure onto a BIM model allows verification of construction accuracy by checking for discrepancies. Comparing the model with hidden components such as piping or rebar can also facilitate early detection of construction errors. Moreover, integrating point clouds into BIM models simplifies planning for renovations or expansions. With coordinates aligned at RTK-level precision, point cloud data essentially becomes a "measured BIM model," greatly contributing to the creation of digital twins.

​

Map Creation and Public Surveying through GIS Integration

In the surveying field, the use of 3D point cloud data for creating and managing maps for national and local governments is becoming increasingly widespread.

Public surveying using drones must adhere to government-established standards for acquiring high-precision point cloud data. For example, the "Manual (Draft) for Public Surveying Using UAVs" specifies positional accuracy categories for 3D point clouds at within 5 cm, 10 cm, or 20 cm.

Using RTK-equipped UAVs or GNSS surveying devices allows users to quickly survey extensive areas while meeting stringent accuracy requirements. The resulting high-precision point clouds can be imported into GIS software to generate contour maps, digital terrain models (DTM), and serve as foundational data for creating detailed 3D maps of roads and rivers.

In general surveying applications, especially when setting up fixed base stations is challenging, network RTK using Ntrip has become increasingly popular, enabling direct field capture of georeferenced point clouds. Portable RTK positioning devices such as LRTK are particularly valuable for surveys demanding high mobility, including forestry assessments and disaster site mapping.

Indeed, even in mountainous or remote locations beyond internet coverage, the LRTK antenna—capable of receiving Japan's Quasi-Zenith Satellite System (QZSS) CLAS signals—reportedly enables real-time global positioning with centimeter-level accuracy.

In this way, combining RTK with point cloud scanning is enabling highly accurate and efficient data acquisition for map creation and surveying in public infrastructure projects.

​

Steps for Utilizing RTK and Point Cloud Scanning

To perform high-precision 3D surveying, selecting appropriate equipment and following accurate procedures are crucial. Below is a basic workflow for conducting surveys by combining RTK positioning and point cloud scanning.

1. Equipment Selection and Preparation: Select suitable equipment according to your measurement target and on-site environment. For example, appropriate combinations include an RTK-enabled drone with aerial photogrammetry for surveying extensive terrain, a handheld laser scanner combined with an RTK receiver for measuring interiors of buildings, or a smartphone paired with an LRTK device for as-built surveys on civil engineering sites. When selecting RTK-GNSS equipment, choose either a standalone setup (base station + rover) or a network-based solution (such as a VRS correction service), and install a base station if required. With a solution like the LRTK Phone, preparation is as simple as attaching a dedicated GNSS receiver to a smartphone and configuring the Ntrip connection.

2. On-Site Coordinate Alignment (Positioning Setup): Upon arriving at the survey site, first verify the reference coordinate system to be used. Determine whether you will align your data to a public coordinate system or a specified local coordinate system, and then configure your RTK system accordingly. If installing a base station, set it up over a known reference point—such as an electronic reference point, a triangulation station, or a previously surveyed location—and initiate positioning. In the case of a network-based RTK system, activate your receiver, connect it to the correction data service, and confirm the acquisition of an RTK FIX solution (integer ambiguity resolved). Ensuring stable positioning at centimeter-level accuracy is crucial; typically, the equipment will indicate this status through labels like "FIX" or a green indicator.

3. Conducting Point Cloud Scanning and Georeferencing: When using laser scanners, begin scanning with device coordinates continuously corrected by RTK. For mobile mapping applications, which involve scanning while moving, combine an IMU (Inertial Measurement Unit) and GNSS to form an INS (Inertial Navigation System), correcting the trajectory while capturing point cloud data. In photogrammetry, each captured photo is geotagged with high-precision coordinates derived from RTK. For example, the LRTK app allows simultaneous photo capture and centimeter-level positioning per image, enabling high-accuracy point cloud creation without requiring post-processing. In some cases, including known targets (such as markers or prisms) in several photographs can facilitate automatic coordinate alignment within the software. The crucial aspect here is embedding an accurate coordinate reference within point clouds or photos, significantly reducing the need for positional corrections in later processing steps.

4. Data Alignment and Accuracy Verification: Once scanning is complete, review the acquired point cloud data and perform any necessary fine-tuning. When integrating multiple scans, use alignment algorithms such as ICP (Iterative Closest Point) to merge the datasets accurately. With RTK positioning, significant misalignments rarely occur, but verifying accuracy with control points (pre-surveyed known points on the ground) is highly recommended. Specifically, evaluate accuracy by comparing the coordinates of validation points set within the measurement area against their corresponding points in the point cloud data. Confirm whether these differences fall within an acceptable range (typically within 5 cm). Validation points are usually established using metal markers or target sheets surveyed beforehand with a total station or high-precision GNSS receiver. Although RTK technology typically provides centimeter-level positional accuracy, minimizing the need for substantial corrections, accuracy verification with control points remains essential for quality assurance purposes.

5. Error Processing and Data Cleaning: Finally, unwanted points and noise in the point cloud data are removed to refine the dataset. Noise or extraneous points—such as people, vehicles, or outliers captured during scanning—can be filtered out using specialized software. For instance, isolated points significantly above the ground level or areas with abnormally low point densities are eliminated. Recently, AI-driven technologies for automatically detecting and removing noise points have advanced, allowing efficient generation of clean point clouds. Additionally, depending on requirements, point cloud data may be converted into mesh (polygonal) models, or key features suitable for CAD drawings may be extracted. The resulting high-precision point cloud data can then be directly integrated into desired maps or BIM models, or used for comparative analysis with the original designs.