The Most Commercially Mature Use of Drones

Infrastructure inspection is where drone work stopped being a pilot program and became a line item. The reason is simple arithmetic. A bridge inspection using a snooper truck costs $8,000 to $25,000 per span, requires lane closures, and books weeks in advance. A cell tower climb costs $1,500 to $3,000, requires two certified climbers, and carries a fatality rate that makes tower climbing one of the more dangerous occupations in the country. A drone does the visual portion of both jobs in a fraction of the time, with nobody exposed and no traffic control.

That does not make the drone a replacement for the engineer. It replaces the access method. The inspection judgment still belongs to a licensed professional, and any product sold into this market has to respect that division or it will not be accepted by the agencies that sign the reports.

What Gets Inspected and With What Sensor

Bridges. Concrete spalling, delamination, exposed rebar, corrosion at bearings and expansion joints, and cracks measured in tenths of a millimeter. High-resolution visible imagery is the primary tool, and crack width measurement requires a known ground sample distance, meaning the camera-to-surface distance has to be controlled and recorded. Thermal imaging finds subsurface delamination in concrete decks because a debonded layer heats and cools at a different rate than sound material, and the useful window for that survey is a few hours after solar loading.

Cell towers. Antenna azimuth and downtilt verification, mount condition, coax and jumper routing, grounding, and lighting systems. The deliverable is often a dimensioned 3D model rather than photos, because the carrier wants to plan a new equipment install without a second climb.

Roofs. Thermal survey after sunset finds trapped moisture in insulation, which retains heat and shows as a warm plume against a cooled membrane. This is one of the highest-value and lowest-difficulty applications in the entire category, and it is the one that funds a lot of small inspection businesses.

Transmission and distribution lines. Corona discharge, damaged conductors, hot connections found with thermal, insulator condition, and vegetation encroachment measured from lidar returns. Utilities have both the budget and the regulatory pressure to fund this at scale.

Sensor selection follows the defect, not the other way around. Thermal payload integration is covered in choosing a thermal camera payload, geometry capture from imagery in drone photogrammetry, and direct range measurement in lidar drone mapping. For most structural work, photogrammetry at high overlap gives better texture for crack detection, while lidar wins under vegetation and on thin linear features like conductors and lattice members.

The Repeat Survey Is Where the Value Lives

A single inspection tells you the condition today. A second inspection flown along the identical path six months later tells you the rate of change, and rate of change is what drives maintenance budgets and risk decisions. This is the difference between selling a service and selling a monitoring program, and it is worth an order of magnitude more.

Getting there is an engineering requirement, not a software feature. Repeatability demands the aircraft fly the same waypoints at the same altitude with the same camera settings and the same sun conditions, and that the resulting datasets register to each other within a tolerance tighter than the defect growth you are trying to detect. That usually means RTK positioning, permanent ground control targets on the structure, and a mission plan stored and replayed rather than re-flown by hand.

Why the Flying Is Hard

  • GPS is unreliable exactly where you need it. Under a bridge deck, inside a lattice tower, or between transmission structures, satellite geometry degrades and multipath from steel produces confident but wrong positions. Visual-inertial navigation and terrain-relative positioning are the practical answer, covered in GPS-denied drone navigation.
  • Close proximity, hard surfaces. Useful inspection distances are 3 to 15 ft from concrete or steel. Ground effect and ceiling effect near large surfaces destabilize the aircraft, and prop wash off a flat face pushes it back unpredictably. Reliable close-quarters flight requires obstacle sensing that works against featureless gray concrete, which many stereo systems do not. See obstacle avoidance sensors and algorithms.
  • Electromagnetic interference. Energized transmission lines and active antenna arrays interfere with compasses, GPS, and control links. Test the aircraft in the real environment before promising performance, and provide a documented minimum approach distance.
  • Under-deck inspection. Flying inverted-looking cameras beneath a bridge means no GPS, no sky, and no downward optical flow reference. This is where purpose-built inspection aircraft with caged rotors and upward-facing sensors earn their price.
  • Line of sight. Long linear assets and large facilities push past what an observer can see, requiring the waiver process described in BVLOS operations.

From Flight to Deliverable

Nobody buys images. The client buys a report their asset management system can ingest and their engineer can sign. That means every defect tagged with a position on the structure, sized in real units, classified against the client's own condition rating scale, and linked back to the source frame for audit. Automated defect detection using computer vision is now good enough to pre-screen a dataset and flag candidate cracks and corrosion, which cuts review labor substantially, but it must present as a suggestion to a human reviewer rather than a conclusion. A ten thousand image dataset with no triage layer is not a deliverable, it is a burden shifted onto the customer.

The output format question is usually decided by the client's existing software, so ask which platform they use before designing an export.

Who Pays, and the Next Operating Model

State departments of transportation, bridge owners, tower REITs and carriers, electric utilities, commercial roofing contractors, industrial plants, and insurance carriers all fund this work, and each has a different procurement path. Utilities and DOTs tend to buy multi-year programs; roofing contractors buy equipment outright.

The direction of travel is toward permanently installed systems: an aircraft living in a weatherproof enclosure at the site, flying a scheduled route without a crew, and uploading results automatically. That converts a service call into a subscription and is the model most infrastructure owners are moving toward. The enclosure engineering behind it is covered in drone-in-a-box docking stations.

Build an Inspection System That Produces Signable Reports

Projects House develops inspection UAS and the data pipeline behind them: sensor selection, close-quarters flight performance, repeat-survey planning, and defect reporting in the format your client's asset system expects. Describe the asset class and defect types through our contact form.