What is the latest technology "LRTK" that eliminates AR drift? The accuracy revolution in construction DX

Table of contents

• What is AR drift?

• Main causes of AR drift

• Smartphone GPS errors and their impact on AR

• Misalignment due to sensor accuracy and environmental factors

• Ways to eliminate AR drift

• What is high-precision positioning with RTK?

• Eliminating smartphone GPS errors with LRTK

• Simple surveying with LRTK

• Summary

• FAQ

What is AR drift?

In recent years, AR (augmented reality) apps using smartphones have increased, allowing virtual objects and navigation information to be displayed in the real world. However, many users still suffer from misalignment of AR displays known as “AR drift.” For example, have you ever seen arrows in an AR navigation feature on a map app displayed off the actual road, or experienced characters in an AR game appearing in the wrong position? On construction sites, when CAD models or piping models are displayed in AR, their positions can shift and fail to align correctly. This phenomenon, where virtual objects displayed in AR appear to be offset from their real-world positions, is called “AR drift (AR display misalignment).” This article explains in detail why AR drift occurs, its main causes, and possible solutions.

Main causes of AR drift

There are several reasons why AR display positions can become misaligned with reality. The main factors include:

• Smartphone GPS position errors – Built-in smartphone GPS can have errors of several meters or more, and is a primary cause of AR object misplacement.

• Limits of smartphone sensor accuracy – Limited accuracy of gyroscopes and electronic compasses (magnetometers) can introduce errors in detecting device orientation and tilt, affecting AR display direction and position.

• App processing and environmental factors – Limitations in an AR app’s alignment algorithms and the usage environment (GPS signal reflection by tall buildings, nearby ferrous materials that disturb magnetism, low-light conditions, etc.) can also cause misalignment.

The combination of these factors can make the AR image on a smartphone screen appear slightly offset from its true position in the real world. Among these, the most significant influence is often the error in the smartphone’s GPS-derived location. Below, we take a closer look at how GPS errors affect AR drift.

Smartphone GPS errors and their impact on AR

The GPS (Global Positioning System) used by smartphones to determine current position has unavoidable positioning errors. Under ideal conditions, typical built-in smartphone GPS accuracy is said to have horizontal errors of several meters (approximately 5–10 m (16.4-32.8 ft)). In urban areas or canyon-like streets between high-rise buildings, satellite signals can be reflected (multipath) or blocked by buildings, causing errors to reach the teens of meters and in some cases exceed 30 m (98.4 ft). Vertical (height) errors are also large; deviations of around 10 m (32.8 ft) are not uncommon.

These GPS accuracy limitations can be fatal for AR. In alignment-based AR that displays objects in the real world based on location information, if the smartphone’s recognized “current location” is off, then the real-world coordinates where virtual objects should be placed will also be wrong. For example, a virtual object that should be placed on the ground right in front of you may appear to float several meters to the side due to GPS error, or may even be displayed inside a building. Height errors in smartphone GPS can also cause AR objects that should be placed on the ground to appear to float in midair or be submerged below the ground, producing vertical misalignment.

Many AR apps determine the initial placement of AR content based on the current location obtained from the smartphone’s GPS, then use the phone’s accelerometer and gyroscope to track movement and update the display. However, if the initially set position is off by several meters, no amount of subsequent sensor-based tracking will bring the AR display’s absolute position into alignment with reality. In short, smartphone GPS error is often the fundamental cause of AR drift.

Misalignment due to sensor accuracy and environmental factors

Besides smartphone GPS error, there are several other factors that can cause AR display misalignment (AR drift). First, there is the issue of sensor accuracy on the phone. Since AR overlays objects based on camera imagery, accurate knowledge of the phone’s orientation and tilt—provided by gyroscopes, accelerometers, and electronic compasses—is essential. But these sensors are not perfect. For example, an electronic compass (magnetometer) is susceptible to local magnetic fields; if there are nearby transmission towers, large metal objects, or powerful electrical equipment, the compass reading can become skewed. If the compass’s indicated north is off by 5°, the position at 100 m (328.1 ft) ahead will be displaced by about 8–9 m (26.2-29.5 ft). Gyroscopes can also accumulate small drift (zero offset drift) over long usage, which can cause AR objects to slowly shift.

Next are app-side processing and environmental factors. Advanced AR apps now use technologies such as VPS (Visual Positioning System) or AR cloud services that leverage camera feature points to correct self-location. However, if an app does not use such mechanisms, or the surroundings lack identifiable features—such as featureless white walls or dark environments—accurate alignment becomes difficult. If the smartphone GPS is off by several meters and there are no visual cues to correct that error, the app cannot compensate. Additionally, the phone’s processing power limits and algorithmic errors in the app, while typically small, can still affect AR display accuracy.

Overall, AR display misalignment (AR drift) results from the overlap of “position information (GPS) errors” and “sensor/algorithm errors.” So how can we eliminate these misalignments and make AR match reality exactly? The next section looks at concrete solutions.

Ways to eliminate AR drift

To eliminate AR positional errors and display virtual objects accurately in the real world, countermeasures should be taken for each of the causes mentioned above. Major countermeasures include:

• Improve GPS accuracy: The most effective measure is to dramatically improve the accuracy of the smartphone’s position information (GPS accuracy). This can be accomplished using RTK, a positioning technology described below.

• Properly calibrate sensors: Basic measures such as recalibrating a misaligned compass by waving the phone in a figure-eight pattern, or periodically restarting an app to reduce gyroscope drift, can be effective.

• Use feature-point-based position correction: If available in the AR app, use technologies like VPS (Visual Positioning System) or AR cloud services that recognize surrounding buildings or landscapes with the camera to correct self-positioning.

• Choose your environment: Environments that are extremely unfavorable to GPS or compass performance (under high-voltage power lines, in downtown high-rise districts, or in areas with strong magnetic fields) tend to produce larger errors. If possible, moving to a slightly more open area before using AR can improve stability.

Among these, improving position information (GPS) accuracy remains the key. If GPS errors of several meters can be reduced down to a few centimeters, AR display misalignment can be greatly reduced. The trump card for this is RTK (Real Time Kinematic) high-precision positioning technology.

What is high-precision positioning with RTK?

RTK (Real Time Kinematic) is a technique that corrects satellite positioning (such as GPS) errors in real time. Specifically, a known-accurate reference station (base station) and a mobile receiver (rover, in this case a smartphone) compare satellite signal data, and the error information obtained at the base station is sent to the rover. The rover applies that correction information to its own positioning results, greatly reducing satellite positioning errors.

Using RTK can enhance GPS positioning, which normally has errors of several meters, to extremely high accuracy within a few centimeters. Originally used in specialized fields like civil engineering surveying, miniaturization and cost reductions in recent years have made RTK-based high-precision positioning increasingly available on consumer smartphones. For example, in Japan, RTK solutions for smartphones that support the centimeter-level augmentation service (CLAS) provided by the Quasi-Zenith Satellite System “Michibiki” have appeared. This mechanism allows smartphones to directly receive correction information from satellites and perform RTK positioning even in mountainous areas where network connectivity is difficult.

That said, performing RTK positioning on a smartphone by itself is not easy. RTK requires high-sensitivity antennas that support multiple frequency bands, but the GNSS chips and antennas built into smartphones are small and limited in performance. Also, receiving and processing RTK correction data from reference stations (via internet Ntrip streams or satellite CLAS signals) requires apps and settings that can be complex and difficult for non-experts. To address this, solutions have emerged that attach external devices to smartphones so anyone can easily use RTK positioning. This system is called “LRTK.”

Eliminating smartphone GPS errors with LRTK

LRTK is a high-precision GNSS solution used in combination with a smartphone. By attaching a pocket-sized dedicated device (an RTK-capable GNSS receiver) to the smartphone and connecting via Bluetooth or similar, centimeter-level positioning accuracy can be achieved easily. For example, attaching an LRTK receiver to a smartphone and launching a dedicated app can instantly improve the current position—previously off by several meters with the phone’s built-in GPS—to centimeter-level accuracy.

LRTK devices incorporate high-performance multi-band GNSS modules and antennas, capturing not only GPS but also GLONASS, Galileo, and Michibiki (QZSS) signals simultaneously. They receive correction information from reference stations over networks (Ntrip) or directly from Michibiki’s CLAS signal to perform real-time position correction. As a result, they virtually eliminate the errors that were unavoidable with conventional smartphone GPS, allowing the smartphone’s own position to be determined at a stable cm level accuracy (half-inch accuracy).

By leveraging this highly accurate self-position information, AR display misalignment (AR drift) is naturally resolved. A smartphone equipped with LRTK continuously and accurately knows its position. When the device’s previously meter-level erroneous current location is correctly corrected, virtual objects shown in AR will align precisely with their true positions in the real world.

For example, even when displaying 3D design data in AR on site, projecting it based on LRTK’s high-precision coordinates will make the model appear in the exact position overlapped with the ground or structures. As workers walk around and change viewpoints, LRTK prevents self-position errors, avoiding AR objects from wobbling or shifting. In other words, LRTK enables a stable AR experience without positional offsets. It frees users from complicated pre-alignment tasks and the frustration of misaligned displays, allowing AR content to fully deliver its value on-site.

Simple surveying with LRTK

The benefits of LRTK are not limited to improving AR display accuracy. By combining a smartphone with LRTK, anyone can easily perform position-measurement tasks that previously required specialized surveying equipment and skilled technicians. In other words, a smartphone becomes a “simple surveying instrument.”

Specifically, a smartphone equipped with LRTK can acquire coordinates of arbitrary points on site, perform continuous position measurements to record movement paths (tracks), and more. The collected data can be plotted on the smartphone’s map screen or uploaded to the cloud for sharing and analysis with a single touch. It is also possible to tag photos taken on site with high-precision latitude, longitude, altitude, and heading information at that moment and save them. This allows staff back in the office to accurately identify where photos were taken and the camera’s orientation, greatly improving the accuracy and efficiency of site records.

Moreover, LRTK can assist AR-based tasks such as setting out reference points (bore-marking) and verifying as-built conditions. For example, virtual stakes (“AR stakes”) specified in drawings can be displayed and used as guides on site for where to drive stakes. Also, projecting a pre-construction 3D model on site for intuitive sharing with stakeholders becomes straightforward. These tasks traditionally required surveyors to set up transits or GNSS instruments, compare paper drawings on-site, and perform labor-intensive workflows—tasks that can be executed instantly with a single LRTK-enabled smartphone.

Such “simple surveying with LRTK” is expected to be useful across a wide range of fields, from civil engineering and construction sites to mapping, infrastructure management, and even AR games in the entertainment sector. By solving the positional-accuracy bottleneck in various scenarios, LRTK contributes to productivity improvements on site and the creation of new experiences.

Summary

The problem of “AR displays being misaligned” was primarily caused by errors in smartphone GPS. With meter-level GPS errors, virtual objects inevitably appear misaligned with reality. However, by using LRTK, which makes RTK high-precision positioning easy to use on smartphones, these misalignments can be dramatically reduced. When a smartphone can determine its position to centimeter-level accuracy, AR displays can achieve an almost seamless integration with the real world.

Beyond AR, the high-precision positioning enabled by LRTK delivers strong benefits in many applications. Without specialist knowledge, a single smartphone can perform simple surveying, instantly share site information via the cloud, and enable intuitive verification of construction plans via AR—practically realizing construction DX. If you are troubled by AR display drift or want to perform high-precision positioning easily on site, consider trying an LRTK solution.

FAQ

Q: Why is AR display misaligned if I only use my phone’s GPS? A: Because the positional accuracy of built-in smartphone GPS is limited and can incur errors of several meters or more. If virtual objects are placed in the real world using this incorrect current location, the displayed positions will inevitably differ from reality. Moreover, GPS errors compounded by sensor inaccuracy and environmental effects can enlarge the misalignment.

Q: How accurate is smartphone GPS? A: Typical smartphones commonly have horizontal errors of around 5–10 m (16.4-32.8 ft). In poor environments, errors can reach tens of meters. Vertical accuracy is even worse, and errors of 10 m (32.8 ft) or more can occur. Some recent high-end phones support dual-frequency GNSS and have improved accuracy, but they still cannot achieve centimeter-level precision.

Q: How accurate does LRTK make positioning? A: Using LRTK, smartphone positioning accuracy can be improved to a few centimeters in both horizontal and vertical directions. Field measurements have reported horizontal errors of ±1–2 cm (±0.4-0.8 in) and vertical errors on the order of ±a few cm (±a few in). This is many times more accurate than conventional GPS.

Q: Do I need specialized knowledge to use RTK? A: Traditionally, performing RTK positioning required expertise and complex settings. However, LRTK is designed to be usable without specialist knowledge. Just attach the device to your phone and press a button in the dedicated app; it will automatically connect to a correction service and start RTK high-precision positioning. No difficult operations or settings are required.

Q: In what situations can LRTK be used? A: LRTK can be used to improve AR display accuracy, and in various scenarios requiring high-precision positioning: surveying and as-built management on civil and construction sites, high-accuracy geodata collection, field management in agriculture, autonomous driving trials, and more. It is also useful in disaster site documentation and indoor/outdoor position measurement where conventional GPS struggles; with Michibiki’s CLAS, LRTK can enable high-precision positioning in such scenes.

Q: Is LRTK effective for indoor AR as well? A: Since GPS signals usually do not reach indoors, typical indoor AR relies on markers or VPS (camera feature detection) for relative alignment. However, with LRTK’s indoor positioning features, you can maintain the accurate outdoor-acquired position as you enter a building and continue positioning. It is gaining attention as a method useful in places where position is normally easily lost, such as tunnels or under bridges.

Q: How large and heavy are LRTK devices? A: External LRTK GNSS receivers are very compact and lightweight. Thickness is about 1 cm (0.4 in), and weight is around 150 g, so they do not burden a smartphone when attached. They can fit in a pocket and be attached when needed to start high-precision positioning immediately.