9 On-site Problems with Subsurface Pipe AR | How to Troubleshoot When It Doesn't Display, Shifts, or Doesn't Match

Table of contents

• What is subsurface pipe AR?

• Problem 1: AR content does not display

• Problem 2: AR markers do not work and cannot align positions

• Problem 3: Model position shift due to GPS error

• Problem 4: Model does not match due to coordinate system mismatch

• Problem 5: Model orientation shifts due to compass error

• Problem 6: AR display drifts over time

• Problem 7: Small manual alignment errors expand

• Problem 8: Model scale or height does not match reality

• Problem 9: Subsurface pipe data is outdated and differs from the site

• Achieving high-accuracy AR with simple surveying using LRTK

• FAQ

What is subsurface pipe AR?

Subsurface pipe AR is a technology that virtually displays the locations of underground pipes and cables through the camera of a smartphone or tablet. By overlaying pipe routes as lines or models on the screen showing the ground—based on design drawings or survey data—it visualizes underground piping that cannot be seen with the naked eye. In the construction industry, with initiatives like the Ministry of Land, Infrastructure, Transport and Tourism’s i-Construction and the trend toward DX (digital transformation), AR is attracting attention for use in waterworks and sewage construction, inspection of utilities under roads, and pre-checks during land development.

Traditionally, confirming the position of subsurface pipes required excavating the site while referencing drawings and searching for real piping in trial excavations. With AR, you can intuitively grasp the pipe route data acquired in advance at the site, allowing immediate judgments such as “the pipe is buried from here onward.” A significant advantage is reducing the risk of accidentally damaging lifelines and enabling safer work. AR is useful not only for protecting buried objects during construction but also, for example, when updating buried pipes and locating old pipes: projecting past drawing data onto the ground with AR can efficiently identify buried positions.

However, when you actually try to use subsurface pipe AR on site, you may encounter unexpected problems. Virtual piping that should have been displayed is invisible, or positions are shifted and do not match reality. Below are nine common on-site trouble examples for subsurface pipe AR and how to deal with them.

Problem 1: AR content does not display

Problem: You launched the AR app and tried to display the subsurface pipe model, but the crucial model does not appear on the screen at all. Many site personnel are baffled when they believe they loaded the 3D model or line data for the buried pipe, but see nothing when they point the smartphone at the ground—“AR is not displaying…”

Cause: The first likely cause of a model not displaying is that the AR system cannot recognize the surrounding environment. Smartphone AR places virtual objects only after detecting the ground or structures from the camera image. If the ground underfoot is flat asphalt or the camera is pointed at a featureless surface, the device may have too few feature points to reference and initialization of the AR display can fail. Also, if the loaded subsurface pipe model’s position information is far removed from the current location, the model may be rendered at a distant location and not appear in front of you. For example, if drawing data has been registered with coordinates for another site, the model may exist at completely different coordinates than your current location, resulting in a situation where “nothing is displayed.”

Solution: First, slowly move the smartphone to scan the surrounding environment and give the camera enough information to initialize the AR. If the ground has no texture, temporarily place familiar objects such as staff jackets or toolboxes as markers to increase feature points for easier recognition. If it still does not display, check the model data’s position settings—verify whether it matches the site coordinates or whether an anchor (reference point) is set near the current location. In some cases, you may need to perform model alignment at a known point on site (manual alignment, described later). In short, when AR does not appear, suspect “insufficient environment recognition” and “position information inconsistency,” and address the issue by rescanning the environment and checking data settings.

Problem 2: AR markers do not work and cannot align positions

Problem: AR markers or QR codes prepared as references for AR display do not function, preventing model alignment. On outdoor sites you may affix AR markers to the ground or structures to display buried pipes with high accuracy, but the camera does not respond and you cannot place the model in the intended position.

Cause: While marker-based methods are convenient, they tend to be unstable outdoors. Printed markers or QR codes can become undetectable to the camera due to sunlight reflection, dirt, or distance. Markers placed on the ground are especially vulnerable to being stepped on by vehicles or heavy equipment or peeled off by rain and wind, creating a risk of becoming unusable when needed. Also, when using subsurface pipe AR over a wide area, it is impractical to place markers everywhere needed. The coverage area of a single marker is limited, so ensuring reliable functionality across the whole site is difficult.

Solution: If a marker does not respond, first check the marker itself. Clean it if dirty, and replace it if wrinkled or damaged. Shooting distance and angle are also important—don’t get too close to the marker perpendicular to it; maintain an appropriate distance so the camera can focus. If it still fails to be recognized, consider switching to other alignment methods instead of forcing marker use. For example, match the model visually to fixed site reference points (such as existing manhole covers or building corners), or use high-precision GNSS position correction described later. Marker methods are useful but not foolproof, so have backup alignment methods prepared on site.

Problem 3: Model position shift due to GPS error

Problem: The subsurface pipe model can be displayed on the smartphone, but its position is shifted from reality by several meters. The line that should be directly above the underground pipe is shown several meters to the side, leaving site staff wondering, “This is no good for excavation positioning.”

Cause: The GPS accuracy of a typical smartphone or tablet is said to be on the order of 5–10 m (16.4–32.8 ft). You may have experienced your location being shown on the map app at a building beside the road rather than on the road itself. Similarly, using AR with low positioning accuracy causes a large initial placement error of the model. Subsurface pipe AR displays pipeline routes from drawing data in site coordinates, but if the device’s positioning error is large, even if you think you placed the model at the correct coordinates, you will see it shifted by that error amount.

Solution: Introducing high-precision positioning is effective for this issue. Specifically, using correction technologies like RTK-GNSS raises smartphone positioning accuracy to the centimeter level. Recently, attaching an external compact GNSS receiver to the smartphone or using an RTK mode on built-in GNSS in compatible models allows centimeter-level positioning with ease. If high-precision positioning is difficult, manually fine-tune by comparing your current location and the model position on the map, but as a new norm on sites, RTK-based “non-shifting AR” is gradually spreading. In short, positional shifts caused by GPS error can be resolved by improving positioning accuracy through equipment and settings.

Problem 4: Model does not match due to coordinate system mismatch

Problem: The subsurface pipe model displayed in AR does not align with the site overall. While the model itself is visible, the entire pipe route is offset or rotated compared to the actual positions. For example, a pipe that should be buried parallel to the road according to the drawing may appear to cross diagonally in AR, leading to “I overlaid the drawing but it doesn’t match the site…”

Cause: This happens when the coordinate system of the drawing data does not match the site survey coordinates. If design drawings or CAD data were created in a local arbitrary coordinate system (such as a proprietary baseline), discrepancies in translation and rotation arise relative to global coordinates (WGS84) obtained from smartphone GPS or public coordinate systems. If you display the data in AR without transforming it, the origin and orientation will differ from the site, causing the entire model to be mislocated and misrotated. Also, differences in height reference between the drawing and the site can produce vertical mismatches where the model appears to float or be buried.

Solution: A key countermeasure is to perform a coordinate alignment (georeferencing) of the drawing data beforehand. Survey several known site points and adjust corresponding points in the drawing by translating and rotating them to match. Specifically, measure coordinates of reference points in the drawing (for example, manholes or boundary stakes), calculate the offset for the entire model, and apply that transformation to the data. Doing this yields a model that matches the site coordinate system. For vertical alignment, set the model to the site’s elevation reference to prevent floating or sinking. Although coordinate system inconsistencies may seem complex, the essential point is not to neglect the step of aligning to site references.

Problem 5: Model orientation shifts due to compass error

Problem: The orientation of the subsurface pipe displayed in AR is incorrect. A pipe that should run straight along the road may appear slightly angled in the view, so while the position is roughly correct, the direction differs and you end up needing to “tilt the device a bit to make it match” for fine adjustments.

Cause: The error likely comes from the smartphone or tablet’s orientation sensor (electronic compass). Sites often contain metal objects such as steel-frame buildings, heavy machinery, and guardrails that easily disturb magnetic sensors. If magnetic fields are disrupted by high-voltage lines or vehicles, the device’s north reference can shift, causing the model’s orientation to rotate accordingly. For long buried pipelines, even a slight directional error can lead to a significant apparent positional offset, so compass errors cannot be ignored.

Solution: Properly calibrating the device’s electronic compass can greatly improve the situation. When launching the AR app, perform sensor calibration by moving the smartphone in an “8” pattern as prompted at startup to correct the sensor. Additionally, if strong magnetic materials are nearby, move a bit away before using AR, or compare the model’s orientation with on-ground linear features (such as curbs or building walls) and adjust the model direction accordingly. As a more fundamental alternative to relying on the compass, you can use orientation measurement with a dual-frequency GNSS antenna or align the model using a known site direction. In any case, if you suspect compass error, first try recalibrating the sensor and checking the surroundings.

Problem 6: AR display drifts over time

Problem: The subsurface pipe AR display that initially overlapped correctly gradually shifts while work continues. Reports include experiences where a line that matched at the start appears to float a few tens of centimeters away after walking around for a while.

Cause: Drift results from sensor errors in the device and limitations in the AR engine’s tracking accuracy. A smartphone’s gyroscope and accelerometer accumulate slight errors during prolonged use, affecting self-position estimation. Also, changes in the environment captured by the camera (lighting variations or moving objects) can disturb feature-point tracking and cause the virtual object to slip. Once drift begins, the error accumulates until it is reset, causing the model to gradually drift over time.

Solution: If you feel drift, a quick countermeasure is to reset the AR and re-place the model. Re-aligning the reference at intervals refreshes accumulated error. For example, when you reach a predetermined site point (a known point), reset the AR display and re-align the model there. To reduce drift occurrence itself, use high-precision GNSS or VPS (visual positioning service). Continuously correcting device position with GNSS or matching camera imagery against a pre-mapped database using VPS features provided by some AR platforms helps suppress self-position errors. However, since app and environment support may be limited, it is practical to combine periodic resets with these measures on site.

Problem 7: Small manual alignment errors expand

Problem: A small misalignment during manual initial placement of the AR display expands with distance. For instance, you align the virtual subsurface pipe model to a manhole at the start point, but 100 m (328.1 ft) ahead a discrepancy of tens of centimeters (tens of inches) appears.

Cause: Manual alignment inevitably has subjective elements, and even if you think you aligned it perfectly, a few centimeters of error can remain. That error accumulates and appears larger visually the farther away you go. The longer the buried pipe, the more a slight initial angular error manifests as a large lateral offset at distance. Differences in device posture (height or tilt) also change the perceived alignment, making errors more apparent at long range.

Solution: Use as objective a reference as possible for initial alignment. When reference structures are available, verify angle and position at multiple points to fine-tune the model. Checking two or more points rather than a single point allows more accurate alignment. If errors still occur, set intermediate re-adjustment points. For example, for long pipelines, reset the AR display at a mid-route bend or another manhole and re-align the model there to reset the error. Fundamentally, using high-precision positioning for automatic alignment (such as LRTK methods described later) reduces human alignment mistakes.

Problem 8: Model scale or height does not match reality

Problem: The size or height of the subsurface pipe model shown in AR differs from reality. The pipe’s diameter or shape looks unnatural compared to the real object, or the model appears to float above the ground. Even if position aligns, a wrong scale causes a strong sense of inconsistency on site and may hinder safety checks.

Cause: A likely cause is a mismatch in model scale or units. During CAD conversion, meters and feet may be mistaken, or a model created from a 2D drawing may not be at actual scale, resulting in AR showing a scaled version. Vertically, even if the planar position matches, differing elevation references can make the model appear at an incorrect height relative to the ground. Smartphone AR tends to produce height-direction errors, especially for distant objects, making height-matching along terrain difficult.

Solution: Prevent scale mismatch by standardizing units during data conversion. When preparing drawings or models for AR, confirm that they are at real-world scale—for example, `1 unit = 1 meter`—before use. If the display appears too small or too large, check the original data’s dimension values and adjust them as needed. For height mismatches, set a height offset for the model based on site ground elevation or known point elevations so that “lines that should match the ground in AR” actually overlay the surface. Newer smartphones with LiDAR scanners can capture terrain and help the model conform to the ground. The key is to verify and adjust scale and height so models display at true size and correct elevation.

Problem 9: Subsurface pipe data is outdated and differs from the site

Problem: The subsurface pipe data used for AR does not match actual site conditions, making the display untrustworthy. For example, projecting pipe locations from an old ledger map or past construction drawing may show a pipe where none exists because the route was altered.

Cause: When subsurface pipe data is outdated, renovations or replacements on site can cause discrepancies between drawings and reality. Older buried pipe maps, especially those created decades ago, may also be inaccurate in recording and even if coordinates match, the documented information itself may be wrong. AR visualizes whatever input data you provide, so low reliability of source data leads to “AR display differs from the site.”

Solution: Always use up-to-date, accurate data for AR. If uncertain, confirm actual pipe positions with trial excavation or detection radar beforehand and create data reflecting that information. Recently, methods have increased for 3D scanning exposed subsurface pipes with smartphones or drones during excavation and obtaining precise position models from point clouds. Using such scan technologies lets you digitally record the site as-is and use that for AR without relying on old drawings. In short, careful data preparation for AR is essential; ensuring information matches the site beforehand prevents many troubles.

Achieving high-accuracy AR with simple surveying using LRTK

Above we introduced nine common on-site problems and countermeasures for subsurface pipe AR. If you want to fundamentally eliminate shifts and mismatches, a highly effective approach is to drastically improve positioning accuracy and data consistency. One increasingly notable method is a simple surveying approach using smartphone positioning technology called LRTK.

LRTK combines high-precision GNSS (RTK) and smartphone AR to realize virtually “non-shifting” AR on site. By correcting position to the centimeter level with a small RTK-GNSS receiver attached to the smartphone and using a dedicated app to align design data with site coordinates, you can display design lines that precisely coincide with actual positions when viewing buried pipes in the field. This eliminates the need for complex marker placement or manual alignment, and the pipe AR remains fixed to the correct ground location even as the user walks around.

Additionally, LRTK can use the smartphone’s built-in LiDAR or camera to acquire simple 3D data of the site and immediately reflect it in the AR display. For example, if you scan an exposed buried pipe during a trial excavation and save the point-cloud model to the cloud, you can project its exact position in AR even after backfilling. In this way, LRTK’s simple surveying allows anyone to easily “measure the site and import it into virtual space,” bringing high-accuracy AR overlay—previously hindered by error and effort—within easy reach. If your site is struggling with AR shifts or display issues, it is worth considering such cutting-edge technology.

FAQ

Q1. What equipment and apps are required for subsurface pipe AR? A. Basically, you need an AR-capable smartphone or tablet, an RTK-GNSS receiver capable of positioning with centimeter-level accuracy (half-inch accuracy), and a dedicated app that links them to display subsurface pipe data (e.g., an LRTK app). AR display itself is possible with a standalone smartphone, but RTK-based positioning augmentation is essential to achieve high-accuracy alignment. Recently, compact GNSS units that attach to smartphones have appeared; using one can turn your device into a high-precision surveying and AR tool.

Q2. I only have paper drawings or PDF data. Can I still display subsurface pipes in AR? A. Even if the source design data is paper or PDF, you can make AR display possible with some effort. The recommended approach is converting PDF drawings into CAD data (DXF/DWG). If you can vectorize the line data, you can overlay the pipe routes directly in AR. If CAD conversion is difficult, you can import the PDF as an image and display a simple plan view (scale adjustment will be necessary). The important point is to prepare data that matches site coordinates. Check with the client to obtain original electronic data if possible; if not, at minimum georeference the scanned image before using it in AR. LRTK can directly handle common CAD formats like DXF/DWG, so digitalizing data early is ideal.

Q3. How trustworthy is the accuracy of subsurface pipe AR display? A. With proper preparation, planar position accuracy can be kept within a few centimeters (a few inches). Especially when using RTK positioning and correctly aligning drawing data with coordinates, you can overlay virtual models with accuracy comparable to conventional surveying. Note that smartphone AR tends to produce height-direction errors depending on device posture and environment; distant models may appear exaggerated in elevation differences relative to terrain, so be careful when matching heights on flat surfaces. Overall, with appropriate methods, you can achieve accuracy suitable for stake-out and pipe position setting, making AR a useful on-site tool.

Q4. Can AR visualize buried objects in places where GNSS signals do not reach (indoors or inside tunnels)? A. In environments without GPS satellite reception, achieving outdoor-level precision with RTK is difficult. However, certain methods allow some AR overlay indoors. For example, LRTK has an indoor mode that estimates position using only the smartphone camera and inertial sensors for a short time. Because errors accumulate with time and distance, for long tunnels or large indoor spaces it is recommended to periodically reset position at known points or use QR-code markers in combination to correct drift. In short, although you cannot achieve outdoor-level precision, with ingenuity you can still use AR indoors or underground.

Q5. Will AR completely eliminate the effort of locating buried pipes? A. AR is a visualization aid and not a complete replacement for finding the actual object. Final confirmation of exact buried positions still requires trial excavation with a shovel or verification using detection equipment (metal detectors or ground-penetrating radar). That said, AR’s ability to visualize approximate locations and routes greatly reduces unnecessary excavation and significantly lowers the risk of hitting pipes. Especially with high-accuracy AR like LRTK, you can nearly complete stake-out or hole positioning on the spot, dramatically improving work efficiency. In other words, AR is not a magical replacement for all manual tasks, but when used properly it is a powerful support tool that substantially reduces on-site burden.