Underground Construction Monitoring Drone Guide

Underground construction presents some of the most difficult monitoring conditions in the infrastructure industry. Tunnel excavation, underground stations, utility corridors, shafts, caverns, mines and subterranean facilities are often confined, poorly illuminated, dusty and continuously changing. Satellite positioning may be completely unavailable, communications can be unreliable, and sending personnel into active construction zones can involve additional safety procedures and disruption to ongoing work.

Drones are increasingly useful in these environments because they can provide rapid visual and three-dimensional information without requiring personnel to physically access every part of the site. Equipped with technologies such as SLAM LiDAR, RGB cameras, thermal cameras, lighting, gas sensors and specialised navigation systems, drones can document excavation progress, map underground structures, inspect difficult-to-access areas and create repeatable records of construction activity.

The greatest value comes from treating the drone as a data-collection platform rather than simply a flying camera. Information collected underground can be integrated with CAD, BIM, GIS, point-cloud and digital-twin platforms, allowing construction teams to compare actual site conditions with design information and previous surveys.

However, underground drone monitoring has important limitations. GNSS is normally unavailable, dust can interfere with cameras and LiDAR, repetitive tunnel geometry can challenge localisation, communications can be blocked by rock and concrete, and a visually complete drone survey does not confirm structural safety. Drone information should therefore complement professional surveying, engineering, geotechnical monitoring and construction inspection rather than replace them.

Why Use Drones for Underground Construction?

Traditional underground construction monitoring often requires surveyors, engineers or inspection personnel to enter active work areas. Access may need to be coordinated around excavation equipment, ventilation, blasting schedules, construction vehicles or other operations. Some locations may require scaffolding, elevated platforms or specialist confined-space procedures.

Drones can provide an additional method of collecting information from these areas. A drone can travel through a tunnel, shaft or cavern while recording imagery and three-dimensional measurements. Large areas can potentially be documented quickly, and the same route can be repeated periodically to create a historical record of construction progress.

This can be particularly valuable in locations that are difficult to observe from ground level. Tunnel crowns, high cavern ceilings, ventilation structures and elevated services can potentially be inspected without installing temporary access equipment. The objective is not to remove engineers or surveyors from the process, but to give them better information before deciding where physical inspection is necessary.

Tunnel Construction Monitoring

Tunnel construction is one of the strongest applications for underground drones. Whether a project uses tunnel boring machines, drill-and-blast excavation or other construction methods, the geometry and condition of the tunnel change continuously as work progresses.

Drones can document completed sections, construction areas and installed infrastructure. RGB imagery provides a visual record, while LiDAR can create three-dimensional measurements of accessible surfaces. Repeat missions allow project teams to compare conditions between different dates.

This information can support progress reporting, engineering documentation and coordination between contractors. However, a drone-derived model should not automatically be treated as a replacement for formal tunnel surveying. Where contractual dimensions or engineering tolerances are involved, the required survey accuracy should be independently verified.

Excavation Progress Monitoring

Understanding how much material has been excavated and how closely the excavation follows the planned geometry is important on underground projects. Drone LiDAR can create point clouds representing exposed tunnel or cavern surfaces.

These datasets can be compared with previous surveys or design models. Areas where excavation has progressed become immediately visible in three dimensions. Volume calculations may also help project teams understand material removal.

However, volume calculations depend on the quality of the point cloud and the reference surfaces being compared. Occlusions, dust and positioning drift can introduce uncertainty. Measurements intended for contractual or engineering purposes should therefore follow appropriate survey-control and verification procedures.

Overbreak and Underbreak Assessment

Tunnel excavation does not always follow the exact design profile. Overbreak occurs where more material has been removed than planned, while underbreak describes areas where excavation has not reached the intended profile.

A sufficiently accurate LiDAR point cloud can be compared with the designed tunnel geometry to identify candidate differences. Colour-coded deviation maps can make these areas easier for engineers to review.

This can provide useful information for construction control and planning subsequent work. However, the reliability of the analysis depends on georeferencing and measurement accuracy. A small apparent deviation may result from positioning or registration error rather than actual excavation. Professional survey verification is therefore important where tight tolerances are involved.

Underground Caverns

Large underground caverns are used for transportation infrastructure, hydroelectric facilities, storage, utilities and other major engineering projects. Their size can make complete inspection from ground level difficult.

A drone can fly through the cavern and collect LiDAR and RGB data from walls, ceilings and structural features. This can create a detailed three-dimensional record of the space.

The ability to approach elevated areas can be particularly useful. However, safe stand-off should always be maintained from potentially unstable rock or construction elements. Drone access reduces the need for people to approach some areas, but it does not confirm that the surrounding structure is stable.

Shaft Construction

Vertical shafts create particular challenges for conventional inspection. Depending on the project, shafts may extend tens or hundreds of metres underground and contain temporary platforms, utilities, cables and construction equipment.

Specialised drones can potentially descend through these spaces while collecting imagery or LiDAR. The resulting data can help document shaft geometry and construction progress.

However, vertical environments can be difficult for both flight and localisation. Repetitive shaft geometry may provide limited distinctive information for SLAM systems, while air movement can affect aircraft stability. Communication links can also weaken rapidly with depth. Missions therefore require platforms specifically suited to confined underground operation.

Underground Stations

Metro and railway station construction involves large underground spaces containing platforms, tunnels, service rooms, escalator shafts and extensive mechanical and electrical infrastructure.

Drone mapping can provide regular documentation as these spaces develop. Early missions may focus on excavation and structural geometry, while later missions can record installed services and architectural construction.

Point clouds can be compared with BIM models to support coordination. RGB imagery provides visual evidence of construction progress. The same dataset may therefore support engineering, project management and stakeholder reporting.

However, temporary construction materials can create clutter and occlusion. Repeat monitoring should distinguish permanent structures from temporary equipment.

Utility Tunnels

Utility tunnels may contain water, electricity, telecommunications, heating, cooling or other infrastructure. During construction, drones can document the tunnel geometry and later record the installation of services.

LiDAR provides spatial information about pipes, cable trays, supports and structural elements. RGB cameras provide visual documentation.

This can contribute to as-built records and digital twins. However, the drone can only measure visible surfaces. Pipes or cables hidden behind walls, floors or other infrastructure will not be represented unless they were documented before being covered.

Frequent monitoring during construction can therefore be particularly valuable because it captures information that may later become inaccessible.

SLAM LiDAR for Underground Construction

GNSS signals generally do not reach underground construction sites, making conventional GNSS-based drone mapping difficult. SLAM LiDAR provides an important alternative.

SLAM, or Simultaneous Localization and Mapping, allows the drone or payload to estimate its movement by comparing current LiDAR observations with previously observed geometry. The system builds a map while simultaneously estimating its position within that map.

This makes SLAM particularly valuable for tunnels, shafts and caverns. However, SLAM positioning can drift over distance. Long tunnels with repetitive geometry are particularly challenging because one section may look similar to another.

Survey control, loop closures and careful trajectory planning can therefore improve the final result. SLAM is a powerful localisation technology, but it should not automatically be assumed to provide survey-grade absolute coordinates.

Loop Closure and Long Underground Routes

As a SLAM-equipped drone travels farther from its starting point, small localisation errors can accumulate. One way to reduce this problem is through loop closure.

If the drone returns to an area it has previously mapped, the SLAM software can recognise the same geometry and use it to correct accumulated trajectory error. Underground missions can therefore benefit from routes that deliberately reconnect with previously mapped areas where the site layout permits.

Long linear tunnels provide fewer opportunities for loop closure. In these situations, known survey-control points or other external references may be needed to maintain global accuracy.

The quality of an underground point cloud should therefore be assessed using independent measurements rather than judged only by its visual appearance.

RGB Imaging Underground

High-resolution RGB cameras remain one of the most useful underground monitoring sensors. They can document excavation surfaces, installed infrastructure, equipment and visible construction conditions.

The main challenge is lighting. Underground construction sites may have artificial lighting, but illumination is often uneven. Shadows and dark areas can reduce image quality.

Drone-mounted lighting can improve results. Lighting should be positioned to minimise glare and reflections, particularly around wet surfaces and metallic infrastructure.

RGB imagery can provide valuable visual evidence, but it should not be used to infer conditions that cannot be seen. A surface that appears intact in an image is not automatically structurally sound.

Thermal Imaging

Thermal cameras can complement RGB and LiDAR sensors in some underground construction environments. They measure infrared radiation associated with surface temperature and can reveal temperature differences that may warrant further investigation.

Thermal imaging may assist with identifying candidate moisture-related patterns, electrical heating, mechanical equipment temperatures or differences around ventilation systems.

However, thermal anomalies are not diagnoses. Temperature patterns can result from many environmental and operational factors. A thermal difference does not by itself prove water ingress, electrical failure or structural deterioration.

Thermal data should therefore be interpreted alongside site conditions and, where necessary, investigated by qualified professionals.

Water Ingress Monitoring

Water ingress is an important concern in many tunnels and underground structures. RGB imagery can document visible wet surfaces, dripping water or staining. Thermal imaging may sometimes highlight temperature differences associated with moisture.

Repeat drone surveys can help teams document how visible conditions change over time.

However, the absence of visible or thermal evidence does not prove that water is absent behind a lining or within surrounding ground. Drone sensors primarily observe accessible surfaces.

Where groundwater behaviour is important to engineering decisions, drone observations should be integrated with appropriate geotechnical and hydrological monitoring.

Tunnel Lining Monitoring

Once tunnel lining is installed, drones can provide visual and geometric documentation of accessible surfaces. RGB cameras can record visible joints, surface condition and construction details, while LiDAR provides the geometry of the finished tunnel.

Repeat surveys may help identify visible changes or geometric differences requiring closer examination.

Computer vision can assist by highlighting candidate cracks or surface anomalies in imagery. However, automatic detection should be treated as screening. Image resolution, lighting and viewing angle affect whether a feature can be observed.

Professional inspection remains necessary for assessing the significance of any detected feature.

Concrete Construction

Underground projects often contain extensive reinforced-concrete structures, including station walls, shafts, platforms and tunnel linings. Drones can create a repeatable visual record of these surfaces.

High-resolution images may support documentation of visible cracking, staining, surface defects or incomplete work. LiDAR can provide geometric measurements.

However, standard drone cameras and LiDAR do not reveal reinforcement condition or internal concrete defects. Specialist NDT equipment may be required where internal condition is important.

Visible surface condition and structural condition should therefore remain clearly distinguished.

Geological Documentation

During excavation, exposed rock surfaces can contain valuable geological information. High-resolution drone imagery and LiDAR can create permanent records of areas that may later be covered by lining or other construction.

Three-dimensional models can help geologists understand the spatial relationship between visible features. Imagery may support mapping of fractures and geological boundaries where they can be observed.

However, remote imagery does not replace physical geological investigation. Surface appearance alone cannot determine every subsurface condition.

The value of the drone is in providing extensive, repeatable spatial documentation for professional interpretation.

Construction Progress Documentation

One of the simplest and most valuable uses of underground drones is creating a consistent record of progress. Missions can be conducted at agreed construction milestones or scheduled intervals.

The resulting imagery and point clouds provide evidence of how the site looked on a particular date. Project managers can compare successive surveys to understand what has changed.

This is more informative than relying only on selected photographs because the drone can create a broader spatial record.

Consistent routes and sensor settings make comparison easier. Automated processing can then highlight areas of geometric change between survey dates.

As-Built Surveys

As underground construction progresses, the actual geometry may differ slightly from design. Drone LiDAR can contribute to as-built documentation by recording visible constructed surfaces.

The point cloud can be imported into CAD or BIM platforms and compared with design geometry.

However, the accuracy requirement should be established before the mission. A point cloud suitable for visual documentation may not necessarily meet contractual survey tolerances.

Where formal as-built certification is required, appropriate survey control and independent verification should be incorporated into the workflow.

BIM Integration

Building Information Modelling is increasingly central to major infrastructure projects. Underground drone data can provide an important link between the digital design and physical construction.

LiDAR point clouds can be overlaid with BIM geometry. This allows project teams to identify broad differences between planned and constructed features.

RGB imagery can provide visual context for the comparison.

However, a point cloud is not automatically a BIM model. It represents measured geometry rather than intelligent construction objects. Software and professional modelling are required to convert measured features into structured BIM information.

Scan-to-BIM

Scan-to-BIM workflows use point clouds to create or update building and infrastructure models. Underground drone LiDAR can accelerate the collection stage.

Walls, tunnel profiles, structural elements and installed services can potentially be extracted from the point cloud.

AI may increasingly automate this process by identifying common construction elements.

However, automated extraction can misclassify features. The BIM model should therefore be checked against the underlying point cloud and design information before being relied upon for engineering decisions.

Digital Twins

Underground construction projects can eventually be represented as digital twins combining geometry, asset information, construction records and sensor data.

Drone LiDAR provides a practical method of periodically updating the geometric component.

A digital twin could show how a tunnel or station developed over time, allowing engineers to compare construction stages.

Once the infrastructure becomes operational, the same model may support maintenance and asset management.

However, digital twins should clearly identify when information was collected. A detailed 3D model can look current even when parts of the underlying dataset are months or years old.

Deformation Monitoring

Repeat LiDAR surveys can potentially identify geometric changes between different dates. Surface-to-surface comparison software can highlight differences.

This may support screening for areas requiring further investigation.

However, deformation monitoring can involve extremely small movements. Standard drone SLAM LiDAR may not provide the precision or stability required for critical structural monitoring.

Total stations, convergence monitoring systems, terrestrial laser scanners or dedicated geotechnical instruments may remain more appropriate where millimetre-level movement is important.

Drone LiDAR should therefore be used only where its verified accuracy is appropriate for the expected magnitude of change.

Volume Measurement

Underground excavation generates large quantities of material. Three-dimensional surveys can support volume estimation by comparing surfaces from different stages.

A point cloud can represent the excavated geometry, while design models provide reference surfaces.

However, accurate volumes depend on complete coverage and consistent positioning.

Occluded areas should not be silently interpolated into the calculation without understanding the associated uncertainty.

For commercial or contractual measurements, professional survey procedures remain important.

Material Stockpiles Underground

Some underground construction sites temporarily store excavated material or construction supplies in large caverns or staging areas.

LiDAR drones can measure stockpile geometry without requiring personnel to climb unstable piles.

Volume can then be estimated.

However, mass cannot be determined directly from geometry without appropriate material-density information.

The base surface beneath the pile must also be known or estimated. These assumptions should be documented where quantities are commercially important.

Construction Equipment Monitoring

Drone imagery can provide an overview of large underground work areas containing construction machinery.

This may help project managers document site configuration and equipment locations at the time of the survey.

However, the drone should not operate in close proximity to active machinery unless appropriate procedures have been established.

Moving equipment also creates difficulties for SLAM mapping because the mapping system generally assumes that most surrounding geometry is stationary.

Where possible, detailed mapping missions benefit from relatively controlled site conditions.

Ventilation Infrastructure

Ventilation is critical in underground construction. Large ducts, fans, shafts and ventilation infrastructure can be difficult to inspect from the ground.

Drones may provide visual access to elevated or difficult areas.

LiDAR can record duct and structural geometry.

Thermal cameras may provide additional surface-temperature information around operating equipment.

However, airflow can significantly affect small drones. Strong ventilation may make stable flight difficult or unsafe.

The operating limits of the aircraft should therefore be considered carefully.

Air-Quality Sensor Payloads

Drones can also carry compact gas or air-quality sensors in appropriate underground applications. Potential measurements include oxygen, carbon monoxide, carbon dioxide, particulate matter and selected gases relevant to the construction environment.

A drone can potentially map measurements spatially rather than relying solely on fixed sensors.

However, drone propellers disturb surrounding air. This can affect measurements, particularly if sensors are poorly positioned.

Compact drone sensors should also not automatically be treated as substitutes for certified occupational-safety monitoring systems. Their strongest role may be supplementary spatial screening and investigation.

Dust Monitoring

Excavation, drilling and vehicle movement can generate substantial dust. Air-quality sensors mounted on drones may help identify spatial variations.

However, rotor wash can resuspend or redistribute particles.

LiDAR itself can also be affected by airborne dust because laser pulses may reflect from suspended particles rather than solid surfaces.

Heavy dust can therefore reduce both air-quality measurement reliability and mapping performance.

Where possible, high-resolution LiDAR surveys may benefit from being conducted during periods of reduced dust.

Hazardous Atmospheres

Some underground environments can contain flammable or hazardous gases. This creates an important distinction between sensor capability and aircraft suitability.

A drone carrying a gas sensor is not automatically explosion-proof or intrinsically safe.

Most standard commercial drones contain electrical components capable of producing heat or sparks.

Where explosive atmospheres may exist, equipment selection must follow applicable safety requirements and site procedures. The presence of a gas sensor does not make an otherwise unsuitable aircraft safe for entry.

Confined Spaces

Underground construction frequently creates areas that meet confined-space definitions. Drones can reduce the need for initial human entry into some of these spaces.

A protective-cage drone may inspect a chamber, shaft or tunnel before personnel enter.

This can provide visual and geometric information.

However, drone deployment does not automatically change the legal or safety classification of the space. Entry procedures, atmospheric testing and rescue arrangements remain matters for qualified site personnel.

The drone provides additional situational information.

Protective-Cage Drones

Protective cages are particularly useful underground. They prevent propellers from directly striking walls and allow some specialised drones to tolerate limited contact with surfaces.

This can increase survivability in narrow tunnels and complex structures.

However, cages add weight and can partially obstruct sensors.

LiDAR and camera placement therefore needs to account for the protective structure.

Purpose-designed confined-space platforms generally provide better results than simply adding a cage to a conventional outdoor drone.

Communications Underground

Radio communications can be one of the largest operational limitations underground. Rock, concrete, steel and tunnel geometry can block signals.

A drone may lose direct communication after travelling around corners or deep into a tunnel.

Potential solutions include mesh networks, communication repeaters, strategically positioned access points or autonomous navigation that allows the drone to complete limited tasks without continuous control.

However, communication architecture should be designed around the risk of the operation. Loss of a video feed, telemetry and command link can have very different consequences depending on the environment.

Mesh Communications

Mesh networks can extend communications through underground environments by using multiple nodes to relay data.

Repeaters can be positioned along a tunnel as the operation progresses.

This may allow drones to operate farther from the entrance.

However, communication performance remains site-specific. Tunnel bends, metal infrastructure and construction activity can change radio propagation.

Systems should therefore be tested in representative conditions rather than relying solely on theoretical range.

GNSS-Denied Navigation

Because GNSS is normally unavailable underground, drones require alternative navigation methods.

LiDAR SLAM is one of the most important. Visual-inertial odometry, optical flow, IMUs and other sensors may also contribute.

The strongest systems fuse several sources.

LiDAR provides geometric references, cameras provide visual features and the IMU provides rapid motion information.

No single method is perfect. Dust may affect LiDAR, darkness affects ordinary cameras, and inertial sensors accumulate drift.

Multi-sensor navigation therefore improves resilience.

Optical Flow

Optical-flow sensors estimate movement by observing changes in imagery.

They can support stable flight where GNSS is unavailable.

However, performance depends on visible texture and lighting.

Uniform concrete surfaces, darkness or dust may reduce reliability.

Optical flow should therefore be treated as one component of the navigation system rather than a universal solution for underground operation.

Visual-Inertial Odometry

Visual-inertial odometry combines camera information with IMU measurements.

Distinctive visual features help estimate movement.

This can provide effective indoor navigation.

However, underground construction sites frequently contain difficult visual conditions. Dust, low light, repetitive surfaces and rapidly changing illumination can all reduce camera performance.

LiDAR-inertial systems may provide greater robustness in some environments, particularly when combined with visual sensing.

LiDAR-Inertial Odometry

LiDAR-inertial odometry combines laser measurements with IMU data to estimate movement.

The IMU provides high-rate motion information, while LiDAR geometry corrects accumulated error.

This is particularly valuable underground because it does not depend on visible light.

However, long feature-poor tunnels can still create drift.

Professional mapping therefore benefits from survey control and repeat observations wherever practical.

Autonomous Underground Drones

Advanced drones can increasingly navigate underground without direct manual control for every movement.

The aircraft may build a map, avoid obstacles and follow planned routes.

This can support repeatable construction monitoring.

However, autonomy should include conservative failsafes. The drone needs to respond appropriately when localisation confidence drops, battery becomes low or communications deteriorate.

Autonomous capability does not eliminate operational risk. It changes how that risk is managed.

Repeatable Monitoring Missions

Construction monitoring becomes more valuable when surveys are repeated consistently.

The drone can follow similar routes weekly or at agreed project milestones.

Each dataset becomes another snapshot of the underground environment.

Software can compare the latest survey with earlier data and highlight areas that have changed.

Repeatability also supports better progress reporting because teams are comparing similar datasets rather than unrelated photographs.

AI-Assisted Change Detection

AI and automated geometry processing can accelerate comparison between underground surveys.

Algorithms can identify areas where surfaces have moved, new structures have appeared or construction has progressed.

Computer vision may also highlight candidate surface anomalies in imagery.

However, automated detection should be treated as screening. A geometric change may result from temporary construction equipment, registration differences or measurement noise.

AI can direct engineers toward areas worth reviewing, while professionals determine the significance of the observation.

Construction Progress Dashboards

Drone information can be integrated into project-management dashboards.

A 3D model may show which tunnel sections have been excavated, lined or fitted with services.

Images can be linked to locations within the model.

This creates a spatial record of construction progress.

However, automated percentages of completion should be based on clearly defined criteria. The presence of visible geometry does not necessarily mean that a construction package has passed inspection or contractual acceptance.

GIS Integration

Large underground infrastructure projects often contain extensive geospatial information.

Drone data can be integrated with GIS to connect underground assets with surface infrastructure, property boundaries, utilities and environmental information.

To achieve this, the local underground model needs to be accurately transformed into the project’s coordinate system.

Survey-control points can provide this connection.

Without appropriate georeferencing, a SLAM model may be internally consistent but shifted or rotated relative to the wider project.

Georeferencing Underground Data

A SLAM survey commonly begins in a local coordinate system. If the project requires national or engineering coordinates, the point cloud needs to be connected to known control.

Surveyed reference points can be positioned within the tunnel or at access points.

The point cloud is then transformed to the project coordinate system.

This is especially important when combining underground data with surface BIM, GIS or CAD information.

A visually accurate underground model can still be incorrectly positioned globally if this step is neglected.

Survey Control

Major underground construction projects generally already contain survey-control networks.

Drone LiDAR can use these networks to improve georeferencing and verify accuracy.

Known targets or identifiable features can be incorporated into the point cloud.

This provides an independent reference against which SLAM drift can be assessed.

For long tunnels, periodic control can be particularly important.

The objective is to combine the rapid coverage of drone SLAM with the reliability of professional underground surveying.

Point-Cloud Registration

Multiple drone missions may need to be combined to cover a large construction site.

Overlapping point clouds can be registered using common geometry or survey-control points.

This can create one continuous model.

However, automatic registration can occasionally produce a visually convincing but incorrect result.

Independent reference information is therefore valuable.

For very large projects, a structured network of overlapping surveys is generally preferable to relying on one extremely long SLAM trajectory.

Data Accuracy

Accuracy requirements vary significantly by application.

A visual progress model may tolerate more positional uncertainty than an engineering as-built survey.

The project should therefore define the required accuracy before selecting the drone, payload and workflow.

Sensor specifications alone are insufficient. The complete system needs to be considered, including navigation, trajectory processing, calibration and control.

Independent check measurements provide the strongest evidence of actual performance.

Mapping Versus Engineering Measurement

It is important to distinguish between a detailed map and an engineering measurement.

A drone can generate a visually impressive point cloud while still containing centimetres of drift or local misalignment.

For general progress monitoring, this may be entirely acceptable.

For structural tolerances or contractual quantities, it may not be.

The required accuracy should determine whether drone measurements can be used directly or whether conventional survey verification is required.

Data Quality Control

Underground point clouds should be inspected for common SLAM errors. These may include doubled walls, misaligned surfaces, bent corridors or inconsistent loop closures.

Cross-sections can reveal problems that are difficult to see in a three-dimensional viewer.

Control points provide additional verification.

RGB imagery can also help identify whether unusual geometry represents a real object or mapping artefact.

A point cloud should not be accepted solely because it appears complete.

Dust and LiDAR Performance

Dust is a particularly important issue in underground construction.

LiDAR pulses can reflect from airborne particles, creating unwanted points. Dense dust may also reduce the effective range to walls and other structures.

The drone itself can generate dust through rotor wash.

Mission timing can therefore have a major influence on data quality.

Where possible, detailed mapping may be conducted when excavation or vehicle movement has temporarily reduced and suspended dust has settled.

Water, Mist and Spray

Underground sites may contain dripping water, mist or spray. These droplets can produce LiDAR returns and reduce visibility for cameras.

Wet surfaces can also create strong reflections in RGB imagery.

Processing software may remove some unwanted LiDAR points, but severe conditions can still reduce mapping quality.

Environmental conditions should therefore be documented alongside the survey.

A low-quality section should be identified rather than hidden through excessive interpolation.

Lighting

LiDAR can operate in darkness because it provides its own measurement energy, but RGB cameras need light.

Drone-mounted LED systems can provide illumination.

The positioning of these lights matters. Lights located close to the camera can create glare from dust and moisture.

Multiple angled lights may provide more even illumination.

Lighting should also be sufficient for the camera’s shutter speed so that movement does not create excessive image blur.

Flight Speed

Underground mapping generally benefits from controlled, smooth flight.

Rapid movement can reduce LiDAR scan overlap and increase motion blur in RGB imagery.

Fast rotations can also challenge localisation.

A slower mission may therefore produce better information than attempting to maximise coverage.

The optimum speed depends on the LiDAR scan rate, navigation system, lighting and environment.

Consistent movement is generally more important than raw speed.

Battery Endurance

Protective cages, LiDAR, lighting and onboard computers all add weight and consume energy.

Underground drones may consequently have shorter endurance than conventional mapping aircraft.

Mission planning should divide large sites into manageable sections.

Several overlapping flights may be preferable to pushing one mission close to the aircraft’s battery limit.

Safe recovery margins are especially important where the drone cannot simply land anywhere along its route.

Data Management

Regular underground monitoring can generate very large datasets. LiDAR point clouds, high-resolution imagery and BIM models require significant storage and processing.

Projects should establish a consistent naming and version-control structure.

Every dataset should record the collection date, location, sensor configuration and coordinate reference.

Historical surveys can become extremely valuable during later project stages, particularly when a feature has subsequently been covered or removed.

Cybersecurity and Project Confidentiality

Detailed underground maps may contain sensitive information about transport networks, utilities, industrial facilities or critical infrastructure.

Access should therefore be controlled appropriately.

Cloud-processing services need to be evaluated according to project security requirements.

Encryption, user permissions and data-hosting location may be important.

The drone itself should also be treated as part of the project’s information system rather than simply as a camera platform.

Drone-in-a-Box Underground Monitoring

Future underground construction sites may use permanently installed autonomous drone systems.

A protected docking station could recharge the aircraft between missions. The drone could periodically inspect completed tunnel sections or large caverns.

Because GNSS is unavailable, such systems would depend heavily on SLAM, local positioning and robust communications.

Automated missions could create frequent updates to the construction digital twin.

However, underground Drone-in-a-Box systems would require reliable obstacle management because the environment changes continuously as construction progresses.

Integration with Ground Robots

Drones are not always the best platform for every underground area.

Ground robots can carry heavier sensors and operate for longer periods. They are particularly useful where floors are relatively accessible.

Drones can reach elevated structures, shafts, uneven terrain and areas blocked by obstacles.

The strongest future monitoring systems may therefore combine aerial and ground robots.

Both platforms could contribute data to the same SLAM map and digital twin.

Integration with Terrestrial Laser Scanning

Terrestrial laser scanners can provide extremely accurate stationary measurements.

Drone SLAM provides faster mobile coverage and access to difficult areas.

The technologies are therefore complementary.

A construction team might use terrestrial scanners to establish high-accuracy reference geometry and drones to rapidly collect additional coverage between formal surveys.

Drone point clouds can then be aligned to the higher-accuracy control network.

This hybrid approach can provide both productivity and measurement confidence.

Benefits of Underground Construction Drones

The primary benefit is improved access to difficult areas. Drones can collect visual and geometric information from tunnel crowns, shafts, caverns and other locations that may otherwise require temporary access equipment.

They can also improve the frequency of construction documentation. Instead of relying only on occasional formal surveys, project teams can potentially collect more frequent spatial updates.

Drones may reduce personnel exposure to some hazardous or inconvenient areas, although they do not eliminate site-safety requirements.

Perhaps most importantly, they transform construction monitoring from a collection of isolated photographs into a structured three-dimensional record that can be integrated with the wider digital construction workflow.

Limitations

Underground drones operate in one of the most demanding environments for unmanned aircraft.

GNSS is normally unavailable. Communications may be blocked. Dust, moisture and darkness affect sensors. Tunnel geometry can challenge SLAM. Battery endurance can be limited, and construction activity creates moving obstacles.

The resulting data also has interpretation limitations. A visually intact structure is not proof of structural safety. A thermal anomaly is not automatically a defect. A LiDAR difference is not automatically deformation. A normal gas reading does not guarantee that hazardous conditions are absent elsewhere.

These distinctions are essential when drone information contributes to engineering or safety decisions.

Selecting an Underground Construction Drone System

System selection should begin with the environment rather than the aircraft specification. A large cavern, narrow utility tunnel and vertical shaft have very different requirements.

Important factors include aircraft dimensions, protective cage, LiDAR range and field of view, SLAM performance, lighting, camera resolution, communications, battery endurance, obstacle avoidance, onboard processing and environmental protection.

Sensor integration should also be considered. A purpose-designed system in which LiDAR, IMU, cameras and flight controller share accurate timing and calibration will generally provide stronger results than unrelated components added together.

For survey applications, the ability to integrate control points and export data into standard CAD, BIM and point-cloud workflows is particularly important.

A Typical Underground Construction Monitoring Workflow

A professional monitoring programme might begin with the project team identifying which construction areas require documentation and what level of measurement accuracy is needed. Survey control and previous BIM or CAD information can then be prepared before the drone enters the site.

The aircraft performs a planned mission using SLAM or other GNSS-denied navigation while collecting LiDAR and imagery. Additional thermal or environmental sensors may be used where relevant. Data is checked immediately after the mission to confirm sufficient coverage before the team leaves the area.

The trajectory and point cloud are then processed, registered to project control and compared with previous surveys or design information. AI-assisted tools may highlight candidate changes, but engineers and surveyors review the results.

The workflow can be summarised as:

construction monitoring requirement → site and safety assessment → survey-control preparation → GNSS-denied drone mission → SLAM LiDAR and RGB data collection → optional thermal/environmental sensing → trajectory optimisation → point-cloud registration → accuracy verification → comparison with BIM/CAD/previous survey → AI-assisted change screening → professional engineering and survey review → progress reporting → digital-twin update → repeat monitoring.

The Future of Underground Construction Monitoring

Underground construction is likely to become an important market for autonomous inspection and mapping drones. Improvements in LiDAR, visual-inertial navigation, edge computing and AI are reducing dependence on GNSS and continuous manual piloting.

Future drones may autonomously explore newly excavated sections, compare the geometry with BIM models and identify areas where the constructed profile differs from design. They could automatically create updated point clouds and progress reports before returning to an underground docking station.

Multi-robot systems may extend this further. Drones could inspect tunnel crowns and shafts while ground robots map floors and transport heavier sensors. Fixed environmental sensors could provide continuous gas and air-quality information, while mobile robots investigate areas where changes are detected.

AI will increasingly assist with point-cloud classification, construction-progress recognition and change detection. However, professional oversight will remain essential. An algorithm may identify a candidate crack, geometric deviation or thermal anomaly, but determining its engineering significance requires appropriate expertise.

The long-term direction is therefore not simply toward more drones underground. It is toward integrated robotic construction monitoring, where aerial drones, ground robots, fixed sensors, survey networks, BIM and digital twins continuously contribute information to the same project environment.

Conclusion

Underground construction monitoring is one of the most compelling applications for specialised drones because it combines difficult access, GNSS-denied navigation and a strong requirement for frequent spatial information.

Drones equipped with SLAM LiDAR, RGB cameras, lighting, thermal imaging and appropriate environmental sensors can support tunnel construction, shafts, underground stations, caverns, utility corridors and other subterranean projects. They can document excavation progress, create point clouds, contribute to as-built records, inspect difficult-to-access surfaces and provide regular updates for BIM and digital twins.

Their greatest value is not simply replacing a person taking photographs. It is the ability to create repeatable, spatially referenced datasets that allow construction teams to understand how an underground project is changing over time.

However, underground conditions also expose the limitations of drone technology. GNSS is generally unavailable, SLAM can drift, communications can fail, dust and moisture affect sensors, and a detailed 3D model does not automatically establish structural safety or survey-grade accuracy.

The strongest approach therefore combines specialised underground drones, professional survey control, SLAM and LiDAR mapping, consistent repeat missions, BIM and digital-twin integration, automated change detection and expert engineering interpretation.

As autonomous navigation and robotic inspection continue to improve, drones are likely to become a routine part of underground construction monitoring, providing project teams with faster, safer and more comprehensive information throughout excavation, construction, commissioning and eventual infrastructure maintenance.