Seeing something underwater is relatively straightforward. Establishing exactly what is being observed, where it is, whether it has changed and whether that change matters is considerably harder.
This distinction sits at the centre of modern underwater inspection.
A subsea camera can confirm that corrosion is visible on a structure, but it cannot necessarily determine how much material remains beneath the affected surface. Sonar can reveal the geometry of an object where a camera sees almost nothing, but acoustic imagery may not provide the visual detail needed to classify a defect. An ROV can carry both sensors to the target, but the resulting information is considerably more useful if the vehicle also knows precisely where the observation was made.
Underwater inspection is therefore increasingly a systems problem. Robotic platforms provide access, cameras and sonar provide visibility, positioning establishes location, specialist sensors provide measurements, and software brings the resulting information together.
The more significant development is not simply that each of these technologies is improving. They are also becoming more closely integrated.
Inspection ROVs can now carry several complementary sensors, relate their observations spatially, generate three-dimensional representations and compare new data with previous surveys. Autonomous systems are becoming more capable of executing repeatable inspection routes, while AI is increasingly being applied to the task of finding relevant information within the large datasets those missions produce.
The result is a gradual shift in what underwater inspection can achieve. Instead of simply collecting evidence during an occasional survey, the emerging objective is to establish condition consistently enough to understand how it changes over time.
Whether the task is hull inspection, subsea pipeline inspection, underwater bridge inspection, dam inspection or a wider underwater infrastructure inspection programme, the starting point remains the same: what does the operator actually need to know?
Underwater Inspection as an Engineering Problem
A mission intended to confirm that an object is present has very different requirements from one expected to measure corrosion to engineering tolerances.
Likewise, an operator looking for a tear in an aquaculture net does not need the same instrumentation as an engineer returning to an offshore structure to determine whether a known defect has developed since the previous inspection. This distinction separates inspection from simple observation.
Observation can establish that something looks unusual. Inspection seeks to characterise that condition in a form that can be recorded, located, measured, assessed or compared.
The questions may be straightforward:
What is there? Where is it? What condition is it in?
Emerging technologies enable more demanding inspections to go further:
- How large is the affected area?
- Has it changed?
- Is the deterioration accelerating?
- Does the condition exceed an intervention threshold?
- Can another team return to exactly the same point?
For engineering-led applications, underwater non-destructive testing (NDT) may also be required to provide information that images cannot. Visual evidence may show that corrosion exists; ultrasonic testing can help determine remaining material thickness.
Robotic Platforms for Underwater Inspection
The inspection platform determines where the sensors can go, how precisely they can be controlled and how consistently data can be collected.
That makes the choice of platform more consequential than simply deciding whether an ROV or AUV is required.
ROVs and Close-Range Inspection
Remotely operated vehicles remain a natural choice for detailed underwater inspection because the operator stays directly involved in the mission.
Live control matters when the unexpected happens. If a camera reveals an unusual feature, the pilot can stop, change viewpoint, move closer or bring another sensor to bear. This ability to interrogate a finding in real time remains one of the fundamental advantages of an ROV.
Compact inspection ROVs have also changed the logistics of getting useful sensors underwater. A system that can be transported by a small team and launched from a harbour wall, workboat or accessible section of infrastructure can remove much of the mobilisation associated with larger vessel- and diver-led operations.
That accessibility is becoming increasingly important as inspection teams are asked to cover more complex assets with smaller crews and tighter operational windows. Mira Nagle, Head of Marketing at Oceanbotics, highlights the role ROVs can play in extending inspection capability while reducing the demands placed on operators:
“Remotely operated vehicles serve as a crucial force multiplier, allowing operators to reach deeper and more complex environments safely and efficiently. With advances in autonomous navigation, pilots can now perform high-precision scans with minimal intervention, dramatically boosting field productivity.” – Mira Nagle, Oceanbotics
The distinction is important. ROV autonomy does not necessarily remove the pilot from the inspection. Instead, functions such as assisted navigation, station keeping and repeatable scan patterns can reduce the amount of continuous manual control required, allowing the operator to focus more attention on the inspection task itself.
That can be particularly valuable for hull inspection, underwater bridge inspection, dam inspection and other missions where a vehicle must maintain a consistent relationship with a structure while collecting imagery, sonar or measurement data.
Portability, however, has practical limits: Current and tether drag influence control. Payload affects buoyancy, power and stability. A vehicle carrying only a camera and lights has a different requirement from one expected to hold an ultrasonic probe against a structure or operate imaging sonar, laser measurement and positioning equipment simultaneously.
Vehicle specification therefore needs to be considered alongside sensor specification. An accurate inspection instrument delivers little value if the platform cannot hold it consistently enough to obtain a reliable measurement.
Larger ROVs address many of these limitations through greater power, payload and stability, particularly for offshore operations. The trade-off is increased deployment complexity.
ROV selection is consequently a balance between accessibility and control on one side, and stability, payload capacity, endurance and environmental performance on the other.
AUVs, Autonomy and Repeatable Coverage
Where ROVs excel at close investigation, autonomous underwater vehicles are particularly well suited to systematic coverage.
Long pipelines, subsea cables and extensive seabed areas do not lend themselves to continuous manual piloting. An AUV can follow a planned route, maintain survey geometry and collect data over considerable distances without a physical tether to a surface vessel.
Jeremy Harrison, Business Manager at Teledyne Marine, discusses the rise in uncrewed platforms for pipeline inspection and survey projects saying;
“AUVs and gliders are becoming much more established for seabed survey and inspection work. Our Gavia AUV family – the Gavia, Osprey and SeaRaptor – reflects this trend with different vehicle sizes carrying different payload configurations depending on the depth and scope of the survey.” – Jeremy Harrison, Teledyne Marine
For subsea pipeline inspection, this can allow position, burial, free spans, seabed interaction and other conditions to be assessed along long sections of the asset.
Repeatability is equally important. If an autonomous vehicle can follow comparable routes on successive missions, the resulting datasets become easier to compare.
Autonomy introduces different dependencies, however. The vehicle must navigate reliably and deal with its environment without the continuous intervention available to an ROV pilot. Mission planning therefore has to anticipate more of what the platform is likely to encounter, and that boundary is beginning to move.
Increasing onboard computing and sensor processing create the possibility of vehicles doing more than simply following a predetermined route. An inspection system can potentially recognise an unexpected feature, change its trajectory and collect a closer or higher-resolution dataset before continuing.At that point, the vehicle is not merely automating navigation. It is starting to automate part of the inspection decision.
Resident Underwater Systems
The logical extension is the resident underwater robot.
Conventional inspection generally requires a vehicle and support infrastructure to be mobilised whenever new data is needed. A resident system changes that model by remaining associated with the site, docking and recharging before conducting further missions.
The attraction is not simply avoiding individual vessel trips. Frequent, consistently executed inspection creates a much denser record of asset condition. An area of concern can be revisited without waiting for the next major inspection campaign, allowing operators to begin looking at rates of change rather than isolated snapshots.
Communications are another important part of the resident inspection model. Systems such as MetOcean Telematics’ STREAM range can provide satellite connectivity for remote and autonomous platforms, allowing inspection data and vehicle information to be transmitted without requiring physical recovery.
STREAM+ Low-SWaP Iridium transceiver with two-way connectivity (Image: MetOcean)
This can help operators review findings between site visits and supports the wider move towards more frequent, remotely managed inspection.
This kind of tiered connectivity means a resident system’s data can potentially be reviewed between physical visits, rather than only when the vehicle is recovered or docked.
This underlying operational shift is important. Inspection becomes something that can potentially happen routinely rather than an activity that must be mobilised from scratch each time.
Supporting Platforms, Crawlers, and Divers
Not every robotic inspection needs to be performed by a free-swimming underwater vehicle.
Uncrewed Surface Vessels (USV) can collect bathymetric and sonar data themselves or support subsea platforms by providing communications, positioning references and mission-control functions. As autonomy develops, this creates the possibility of distributed inspection architectures in which surface and subsea robots work together rather than relying on a single underwater vehicle operating from a large crewed vessel.
“There’s a clear and accelerating shift towards uncrewed and remotely operated platforms. Operators are looking for ways to carry out surveys that are safer, more cost-effective and greener. A 12-metre USV doing the same inspection job as a 90-metre offshore survey vessel with up to ten personnel makes a strong case on all three fronts.” – Jeremy Harrison, Teledyne Marine
USVs supporting underwater inspection also need awareness of the environment around the asset. FarSounder’s Argos 3D forward-looking sonar, for example, provides real-time information about the seafloor and water column ahead of the vessel while also generating bathymetric data.
This can support safer approaches to inspection sites while adding environmental context for route planning and repeat missions.
Subsea Crawlers address a different requirement. Where a sensor needs to maintain a consistent relationship with a surface, physical attachment can provide stability that a free-swimming ROV cannot easily match. Hulls, tanks and suitable pipeline surfaces may therefore be inspected using crawlers carrying cameras or contact-based instruments.
Divers remain important. Human judgement and dexterity are valuable where access is practical and direct intervention is required. Underwater cameras, ultrasonic instruments and other NDT equipment can all be diver-operated.
The question is increasingly where human presence adds the greatest value. Depth, current, contamination, confined spaces and long durations may favour robotic access, while divers can be reserved for the parts of an inspection where their skills are genuinely needed.
Underwater Cameras and Optical Imaging
Optical inspection is attractive because its output is intuitive. Engineers and operators can look at an image and understand much of what it shows.
Obtaining that image reliably underwater is less straightforward.
Light loss, suspended particles, water colour, viewing distance and vehicle movement all affect the result. A high-resolution camera specification provides little benefit if visibility prevents a clear image from reaching the sensor.
Where conditions allow, subsea cameras can reveal corrosion, cracking, fouling, damaged coatings, deformation, missing components, debris and general deterioration. The same technology might be used to examine a propeller during a hull inspection, identify damage to an aquaculture net, document cracking around a bridge pier or assess offshore equipment.
The required evidence determines the imaging configuration.
A broad field of view helps with navigation and situational awareness. Close examination may require greater resolution, controlled lighting and a consistent working distance. Inspection intended to support measurement places further demands on camera calibration and platform stability.
Lighting is part of the same problem
In clear water, additional illumination can extend working range and recover colour and detail lost with depth. In turbid environments, simply increasing light intensity can make matters worse by increasing backscatter from suspended particles.
Light position, beam angle and distance from the camera therefore influence image quality alongside the camera itself.
This illustrates a recurring principle in underwater inspection: simple sensor choices become integration questions once they are placed on a vehicle and exposed to real environmental conditions.
Moving from Images to Measurements
Visual evidence becomes more valuable when geometry can be extracted from it.
Imagery from subsea stereo cameras derives depth from multiple viewpoints, while photogrammetry reconstructs three-dimensional information from overlapping images. These approaches can turn conventional imagery into data that can be measured and revisited after the mission.
The inspection methodology changes accordingly: a set of photographs adequate for a condition report may be unsuitable for photogrammetric reconstruction. Image overlap, viewing geometry, illumination and platform stability become part of the data-acquisition requirement.
Underwater Laser Systems provide another route to dimensional information. Simple scaling lasers introduce a known reference into an image, while more advanced laser scanners can create detailed representations of the target surface.
The most useful result is not necessarily the most visually impressive model.
Subsea LiDAR extends this further, using laser pulses to build high-resolution 3D point clouds directly, rather than deriving geometry from overlapping images or triangulated scan lines.
Advanced underwater LiDAR solutions for high-resolution 3D point cloud data capture (Image: Kraken Robotics)
This can support dimensional control and digital twinning of subsea assets to a very high level of detail, with outputs such as mesh surfaces, digital terrain models and volumetric calculations feeding directly into engineering assessment.
Optical approaches remain constrained by visibility, however. Once the water prevents reliable light-based imaging, the inspection problem moves increasingly towards acoustics.
A 3D representation becomes an engineering tool when dimensions can be extracted from it, findings can be related to known positions and successive datasets can be compared.
The same principle applies beyond industrial infrastructure. During the Heroic Age Expedition, Voyis’ Discovery Stereo Inspection Series is being used to capture imagery of Quest, Sir Ernest Shackleton’s final expedition ship, at a depth of 390 metres.
Stereo imaging and photogrammetry are being used to create a measurable 3D representation of the wreck, demonstrating how inspection imagery can remain useful for analysis after the vehicle has left the site.
The application is very different from inspecting a pipeline or offshore structure, but the underlying principle is the same: rather than treating underwater imagery simply as a visual record, spatially accurate imaging can create a measurable digital representation that remains useful long after the vehicle has left the site.
Sonar for Inspection Where Cameras Cannot See
Darkness can usually be addressed with artificial lighting. Turbidity is much harder to overcome.
When suspended material restricts the optical working range, sonar may shift from being a supplementary sensor to an operational necessity.
Acoustic imaging is less immediately intuitive than conventional video, but it can reveal objects and geometry where cameras are severely limited. This is particularly relevant to underwater infrastructure inspection in rivers, harbours, reservoirs and nearshore environments where sediment can reduce visibility to very short distances.
For an ROV pilot, forward-looking sonar can first serve as a navigation sensor. It allows the operator to detect a structure or obstacle before it enters visual range and maintain awareness around complex targets.
Once the vehicle reaches the inspection area, imaging sonar can provide more detailed information.
This is particularly useful during underwater bridge inspection and dam inspection, where foundations, piers, intakes, spillways and accumulated debris may have to be examined in poor visibility. Similar requirements exist beneath vessels and around complex subsea structures.
Sonar and optical cameras are therefore better considered complementary technologies than alternatives: Acoustic sensing can locate the target, establish its geometry and guide a vehicle towards it. Optical imaging can then provide the detailed visual evidence required to identify surface condition when visibility allows.
The more interesting development is the increasingly close integration between them.
If sonar, cameras, navigation and motion data are recorded within a common spatial framework, each sensor can compensate for limitations in the others. Rather than producing separate streams of imagery, the inspection begins to generate a more complete representation of the target.
Jeremy Harrison, Business Manager at Teledyne Marine explains this perfectly in real-world terms saying:
“Sonar is another key part of the picture. Our SeaBat T51-S IDH, and the T50, are widely used to build the high-resolution seabed baseline and post-installation survey data operators need. Resolution matters even more than it used to, simply because cables are physically smaller targets than the pipes this technology was originally developed for. Cameras and lighting remain essential too, particularly during trenching, to give operators direct visual confirmation that a cable is correctly seated before it’s covered.” – Jeremy Harrison, Teledyne Marine
Sonar for Mapping and Measurement
Multibeam systems extend acoustic inspection into broader geometric measurement.
By collecting returns across a swath, they can map structures and surrounding seabed in detail. This makes multibeam sonar useful for examining scour, seabed interaction and dimensional change around infrastructure.
The applications extend from offshore assets to bridge foundations and dam structures, where changes around the supporting bed may be as important as visible condition on the structure itself.
Side-scan sonar serves a different role. Its broader coverage is valuable where the problem is initially to locate objects, screen an area or identify changes along a pipeline or cable route.
In subsea pipeline inspection, for example, a broad acoustic survey may reveal seabed disturbance or an area requiring further investigation. A close-range ROV inspection can then target that feature using imaging sonar, cameras or other sensors.
Synthetic Aperture Sonar (SAS) addresses a limitation inherent to conventional side-scan and multibeam systems: resolution that degrades with range and coverage rate. By combining successive acoustic pings into a single synthesised aperture, SAS can sustain high resolution across a wide swath, rather than trading one for the other.
Kraken Robotics’ KATFISH illustrates the current state of this technology. The towfish is actively stabilised and operates at speeds up to 10 knots, producing real-time SAS imagery at 3 x 3 cm resolution, refined to 2 x 2 cm in post-processing, alongside simultaneous 3D bathymetry. That combination of speed and resolution is significant for pipeline and cable route surveys in particular, where operators need both broad coverage and enough detail to characterise seabed condition or identify small targets in a single pass.
Profiling and other specialist sonar systems provide further options where contour or cross-sectional information is required.
No individual sonar architecture is ideal for every inspection. Range, resolution, coverage and update rate trade against each other.
The useful question is therefore not which sonar is best, but what acoustic information the inspection needs to produce.
Positioning, Navigation and Repeatability
An observation becomes considerably less useful if nobody can determine where it was made. This becomes particularly important as inspection moves from one-off condition checks towards repeated assessment.
GNSS cannot directly position a submerged vehicle, so underwater navigation typically combines acoustic positioning, inertial sensors and other vehicle measurements.
The challenge is not simply establishing where a vehicle is at one moment, but maintaining a sufficiently accurate understanding of its position, orientation and movement throughout the inspection.
Pierre Gazull, Manager, Product Marketing, High Performance Mems Inertial Sensors at Tronics Microsystems, points to inertial sensing as a critical part of that navigation architecture:
“In these GNSS-denied environments, inertial sensors play a fundamental role by providing continuous information on position, orientation and motion. High-stability MEMS accelerometers and gyroscopes enable accurate navigation, vehicle control and survey-grade measurements.” – Pierre Gazull, Tronics Microsystems
That continuity is important because acoustic positioning does not necessarily provide all of the high-rate motion information required to control a vehicle or maintain consistent sensor geometry. Inertial measurements can bridge those gaps, while technologies such as Doppler Velocity Logs and acoustic positioning provide external information that helps constrain accumulated navigation error.
The improving performance of MEMS inertial sensors is also relevant to the wider direction of underwater inspection. Smaller, lower-power navigation components are easier to integrate into compact ROVs, AUVs and autonomous inspection platforms, where size, weight and power budgets can directly influence payload capacity and endurance.
USBL, SBL and LBL systems use acoustic signals to establish underwater position relative to known references. Inertial navigation provides continuity when external updates are unavailable, although drift accumulates over time. Doppler velocity logs can help constrain that drift by measuring movement relative to the seabed or water.
The growing use of smaller inspection ROVs is placing particular pressure on navigation payload size. Alec McGregor, Senior AI Engineer at Advanced Navigation, explains:
“As the subsea sector is moving toward remote offshore operations and seeking overall reductions in costs, vessels, fuel time and carbon footprint, operators are opting for smaller-scale remotely operated vehicles (ROVs) for missions such as pipeline inspection and wind turbine inspection.
“This downsizing directly impacts onboard payload capacity. Critical navigation systems – such as Ultra-Short Baseline (USBL) acoustic positioning and Inertial Navigation Systems (INS) – must deliver uncompromising performance within significantly tighter spatial constraints. As such, modern payload design must prioritise systems in ultra-compact form factors that seamlessly integrate into ROV tooling pods, without disrupting vehicle weight, buoyancy, or hydrodynamics.”
– Alec McGregor, Advanced Navigation
Navigation requirements can also extend beyond the inspection itself.
NOVATECH Submersible beacon for asset locating, tracking and monitoring (Image: MetOcean Telematics)
For autonomous and remotely operated platforms, systems such as MetOcean Telematics’ NOVATECH recovery beacons provide an independent means of locating marine assets after they return to the surface.
This supports the wider operational reliability required when inspection vehicles are deployed remotely or for extended missions.
The important point for inspection is that positioning does more than guide the vehicle. It allows imagery, sonar contacts, and measurements to be associated with a known location.
“There is damage somewhere on the structure” is a fundamentally different outcome from “damage was recorded at this position and can be revisited.”
That difference becomes critical when the objective is to monitor change.
A second inspection only provides a meaningful comparison if the vehicle can return to approximately the same place and collect sufficiently comparable data. Without that consistency, differences in viewpoint, sensor geometry or route may be mistaken for changes in the asset itself.
Techniques such as simultaneous localisation and mapping, alongside the fusion of camera, sonar, inertial and acoustic positioning data, are increasingly helping vehicles maintain a more complete spatial understanding of their surroundings.
Underwater Non-Destructive Testing and Condition Assessment
Images and sonar can reveal a great deal, but many engineering decisions ultimately require a measurement:
- How large is the defect?
- How much material remains?
- Has a feature moved?
- Is a protective system still performing as intended?
These are the questions that bring underwater non-destructive testing into the inspection workflow.
Underwater NDT encompasses techniques used to assess material or structural condition without permanently damaging the asset. The appropriate method depends on both the material and the suspected defect.
Ultrasonic thickness measurement is a straightforward example. A camera may show evidence of corrosion, but ultrasound can provide information about remaining wall thickness. Because the sensor may need controlled contact with the surface, the inspection suddenly places additional demands on vehicle stability, manipulation and access.
Cathodic protection measurements answer a different condition question, while laser measurement and photogrammetry can provide non-contact dimensional information. Other inspections may involve specialist methods for flooded-member detection, leak detection, weld assessment or burial measurement.
There is no universal payload because there is no universal inspection problem.
Subsea pipeline inspection illustrates this particularly well: An initial acoustic survey might establish that the pipeline is exposed or unsupported. Close optical inspection can then document surface condition. If the engineering concern is remaining wall thickness, another NDT technique is required.
Increasingly, operators may seek to collect several of these evidence types during the same deployment.
Combining optical imagery, sonar, dimensional measurement and selected NDT techniques can potentially answer several integrity questions without requiring separate mobilisations for each discipline.
That does not mean carrying every available sensor. Payload, power, stability and data quality still impose practical limits. The objective is integration where it improves the inspection, rather than complexity for its own sake.
From Inspection Data to Condition Intelligence
Collecting inspection data has become relatively easy. Making sense of it remains the harder task.
A single mission may generate several video feeds, sonar imagery, vehicle position, depth, heading, 3D geometry, NDT measurements and operator observations.
If those data streams remain separate, much of their combined value is lost.
An image of a defect is more useful when its location is known. A sonar feature becomes easier to interpret when corresponding video can be retrieved. A measurement gains context when previous inspection data from the same location is immediately available.
EdgeDVR 4U digital video inspection system for subsea structural surveys (Image: DIGITAL EDGE SUBSEA)
This is why data management and sensor fusion are becoming as important as sensor performance.
Digital recording systems can synchronise video with metadata such as position, depth and time, while operator annotations allow areas of interest to be flagged during the mission rather than rediscovered afterwards. Spatial data takes that further.
Photogrammetry, sonar and laser scanning can generate point clouds and 3D representations that allow an asset to be examined after the vehicle has left the water. Engineers can measure features, review the inspection remotely and compare new geometry with previous datasets. This is also where the concept of the digital twin becomes relevant.
Software integration is what turns that raw sonar data into something reviewable in the field rather than only after the mission. FarSounder’s SonaSoft software, for example, can interface with hydrographic packages such as QPS Qinsy, allowing incoming Argos 3D sonar data to be examined through several real-time views: a ping-by-ping raw multibeam display for checking swath width and depth below the transducer, and seafloor detections processed into point clouds and bathymetric grids for immediate review and quality control.
The useful idea is not simply creating a realistic 3D image of an underwater structure. The value comes from creating an evolving digital representation into which inspection findings can be located, measured and compared over time.
A model showing corrosion is useful. A model showing where that corrosion is located, how it relates to other measurements and whether it has increased since the previous inspection begins to support condition assessment.
AI Beyond Defect Detection
The volume of data generated by modern inspection systems creates an obvious role for automated analysis. Machine vision and AI can help identify objects, classify features and highlight anomalies within video, imagery and sonar data. The immediate benefit is reducing the amount of material that specialists need to review manually. Detecting something unusual is only the first step. The more significant opportunity is condition interpretation.
Other sensor data can add further context: Sonar might indicate a geometric change, optical imagery may show surface deterioration and NDT measurements may quantify material condition. Bringing those sources together produces a more complete assessment than any one algorithm operating on an isolated image.
AI should not be viewed as a substitute for engineering judgement. Its strongest role is likely to be prioritising attention: identifying which parts of increasingly large inspection datasets merit closer human assessment.
That places greater importance, not less, on disciplined data collection. Poorly positioned or inconsistent data does not become reliable simply because an AI system analyses it.
Major Underwater Inspection Applications
The same technology stack can support very different sectors, but each application changes the balance between platforms, sensing and measurement.
Subsea Pipeline Inspection
HydroPact and ROV for subsea pipe and cable inspection (Image: Teledyne Marine)
Pipelines introduce both scale and precision.
Long routes favour efficient platforms such as AUVs and wide-area sonar. Surveys may assess position, burial, free spans, seabed interaction, debris and other conditions along the route.
When an anomaly is identified, the inspection can become progressively more targeted. An ROV may obtain close imagery, examine local geometry and deploy ultrasonic or other NDT instruments where material condition needs to be quantified.
Subsea pipeline inspection is therefore often a progression from broad coverage to increasingly detailed characterisation rather than a single sensing task.
Autonomy potentially makes that process more repeatable. Regular surveys can establish not merely that a condition exists, but how quickly it is changing.
Underwater Infrastructure Inspection
Underwater infrastructure inspection spans pipelines, power and communications cables, offshore energy structures, bridge foundations, dams, harbour structures and submerged utilities.
The common challenge is not the type of structure but the need to assess something that cannot be fully examined from above the waterline.
An inspection may begin with sonar or broader mapping before moving to targeted optical imaging and specialist measurement. Accessibility, visibility, geometry and the required level of engineering evidence determine the combination.
For long-term programmes, the objective increasingly extends beyond identifying defects. Operators may need to establish whether scour is progressing, corrosion is accelerating, or structural geometry is changing. That makes repeatable positioning and comparable datasets central to the inspection strategy.
Hull Inspection
Hull inspection presents a different operational geometry.
Hull plating, propellers, rudders, thrusters, sea chests, anodes and other underwater components can often be reached by compact ROVs, crawlers or divers. Inspections may look for physical damage, corrosion, biofouling, entanglement or foreign objects.
Close-range manoeuvrability matters more here than long-distance endurance. The vehicle must maintain a useful viewing geometry while operating around curves, appendages and hydrodynamically complex areas.
Imaging sonar can provide context where visibility is poor, while crawlers can offer a stable platform for detailed surface inspection or contact-based measurements.
Dam Inspection
Dam inspection can involve submerged walls, gates, intakes, spillways and other components that are difficult or hazardous to access directly.
ROVs allow these areas to be examined without necessarily dewatering a structure, while imaging sonar can maintain awareness where reservoir visibility is poor.
Potential concerns include cracking, erosion, debris accumulation, displaced components and deterioration around joints or penetrations.
Repeatability becomes valuable when a known area must be revisited to determine whether the condition is progressing.
Underwater Bridge Inspection
Underwater bridge inspection commonly involves piers, abutments, foundations and the surrounding river or seabed.
Scour is particularly important because changes to the supporting bed may not be visible from the surface. Sonar and bathymetric systems can map the area around foundations, while ROV-mounted cameras can investigate cracking, exposed reinforcement, impact damage and general deterioration.
Rivers introduce their own complications. Current, debris and suspended sediment can restrict vehicle control and optical visibility, increasing the value of sonar.
The inspection may therefore need to assess both the bridge structure and the environment supporting it.
Ports, Aquaculture and Other Applications
Ports and harbours present many similar challenges, with quay walls, pilings, locks and other submerged structures frequently located in turbid water. Compact systems that can be deployed directly from existing infrastructure can make routine inspection possible without major disruption.
Aquaculture places greater emphasis on frequent visual access. Nets, cages, moorings and anchors need regular checks for damage and fouling, making portable ROVs and camera systems particularly useful.
Scientific applications can blur the distinction between inspection, monitoring and survey. The same cameras, sonar, navigation and robotic platforms may be used to revisit instrument installations, seabed experiments or defined environmental sites.
Designing an Effective Inspection Mission
Despite the diversity of applications, effective underwater inspection usually follows the same underlying logic: reduce uncertainty in stages.
The first task is defining what needs to be established.
The target is then located and accessed: Sonar or existing coordinates may provide the starting point, while the choice of vehicle depends on distance, geometry, environment and the level of control required.
Once on site, the platform needs sufficient situational awareness to operate safely around the target. Optical cameras may provide this in clear water; sonar becomes increasingly important as visibility deteriorates.
Inspection can then move from broad observation to specific measurement.
An acoustic survey identifies an anomaly. An ROV approaches it. Optical imaging establishes visible condition. A laser, ultrasonic probe or other NDT sensor quantifies the relevant feature.
The information then needs to be preserved with enough context for it to remain useful. Position, time, imagery, sonar, measurements and operator observations should be connected wherever possible.
When inspection is likely to be repeated, the next mission should also influence the first. Consistent routes, sensor settings and positioning make later comparison significantly more reliable.
The key technology questions include:
- What exactly needs to be inspected?
- Does the mission need to locate, observe, measure or monitor change?
- What level of visual, acoustic or dimensional evidence is required?
- How will depth, current and visibility affect the chosen sensors?
- Is the target localised or does the inspection cover many kilometres?
- Is live operator control important, or is autonomous coverage more useful?
- Does measurement require physical contact?
- Is underwater NDT required?
- How accurately must each observation be positioned?
- Will the same location need to be revisited?
- Can several useful inspection disciplines be combined within the same deployment?
- How will the resulting information support the final engineering, operational or scientific decision?
These questions often lead to a mixed system rather than a single technology choice.
A camera may provide the best final evidence but still require sonar to reach the target. An accurate sensor may need a larger platform simply to remain sufficiently stable. An AUV may provide efficient coverage but still require an ROV to investigate an anomaly. Positioning may appear secondary until the first defect needs to be found again six months later.
Inspection performance is governed as much by integration and methodology as by the specifications of individual components.
From Periodic Inspection to Continuous Condition Awareness
The most capable underwater inspection system is not necessarily the one carrying the greatest number of sensors. It is the system that removes enough uncertainty for the user to make the required decision.
For a routine hull inspection, that may mean a compact ROV, a well-positioned camera and suitable lighting. Underwater bridge inspection or dam inspection in turbid water may require imaging sonar as well as optical sensors. Subsea pipeline inspection may begin with autonomous coverage and broad-area acoustic sensing before moving to a close ROV inspection and underwater NDT.
What links these missions is the relationship between platform, sensor, position and data. Increasingly, intelligence connects those elements as well.
An autonomous platform can collect several complementary datasets. Navigation places those observations within a common spatial framework. Software compares new findings with previous inspections. AI can help determine which areas warrant closer attention.
A more advanced system could then respond to what it detects.
Instead of completing a predetermined route regardless of the findings, the vehicle identifies something unusual, moves closer, collects additional information and then continues the mission. That new evidence can be incorporated into the digital representation of the asset and prioritised for engineering review.
Resident underwater systems extend the idea further: A vehicle associated permanently with an asset could conduct routine inspections, return to a docking station and revisit particular areas when required. Combined with automated analysis and historical data, this begins to look less like a succession of independent surveys and more like an ongoing condition-monitoring system.
Not every underwater inspection requires this degree of autonomy or integration. Much of the sector will continue to rely effectively on relatively straightforward combinations of ROVs, cameras, sonar and experienced operators, but the direction is significant.
Underwater inspection is moving beyond the challenge of simply gaining access to an inaccessible target. The larger opportunity is to collect consistent, spatially referenced evidence that allows condition to be understood and change to be measured.
When robotics, sensing, navigation, measurement and data are designed around the same operational question, underwater inspection becomes more than a view beneath the surface. It becomes a repeatable way of understanding what is there, what condition it is in and, increasingly, what is changing.
