ASSET / BRIDGE
Bridge inspection: getting close to the places nobody can see
Girder soffits, bearings, the underside of the deck: bridges start to fail where people find it hardest to reach. This guide separates what drones can take over from snooper trucks and scaffolding, and what they cannot.
On this page
Where bridges start to deteriorate
Bridge deterioration falls into three families: concrete deterioration (cracking, spalling and exposed rebar, efflorescence, chloride attack, alkali–silica reaction, freeze–thaw), steel deterioration (corrosion, coating breakdown, fatigue cracks, loose or missing bolts), and problems at bearings, expansion joints and drainage. All of them concentrate where water, salt and repeated loads get in.
| Element | Typical defects | If missed |
|---|---|---|
| Deck underside | Cracking, efflorescence, spalling, punch-through | Falling concrete, deck failure |
| Steel girders & cross-beams | Corrosion, coating loss, fatigue cracks | Section loss, crack growth |
| Girder ends & bearings | Leak-driven corrosion, seized bearings | Unintended forces on girders and piers |
| Joints & drainage | Leaks, clogged drains | The source of water and salt for everything below |
| Piers & abutments | Cracking, scour of foundations | Settlement or tilting |
Today’s inspection and its pain points
In Japan, after the 2012 Sasago Tunnel ceiling collapse, periodic inspection of road bridges (and tunnels) every five years, based on close-up visual inspection, became mandatory in July 2014. Inspectors get within arm’s reach and tap or touch where needed. Each bridge is rated in four categories from I (sound) to IV (urgent action).
The means of getting close become the cost and the risk. Under-bridge inspection vehicles (“snoopers”) need lane closures and cannot reach every span. Tall piers and river spans need scaffolding, rope access or even a boat. In the United States routine bridge inspections are required at intervals of up to 24 months (12 months for bridges in poor condition), and owners face the same problem.
| Access method | Strength | Pain point |
|---|---|---|
| Snooper truck | Hands-on, can sound concrete | Lane closures and congestion, traffic risk, limited reach |
| Aerial lift / scaffolding | Reliable close access | Set-up cost and days; hard over rivers and railways |
| Rope access | Tall piers and arches | Few qualified technicians, fall risk |
| Boat | Piers and soffits over water | At the mercy of current and weather |
| Binoculars / zoom | Cheap and fast | Rarely accepted as close inspection |
The other pain is people. Japan has roughly 720,000 road bridges (end of FY2024), many of them owned by municipalities, and many small towns have no bridge engineer at all. Bridges more than 50 years old rose from about 130,000 at the end of FY2018 to about 230,000 at the end of FY2024. The number of ageing bridges grows while the number of inspectors does not. That is why drone inspection is becoming a necessity, not a convenience.
What drones change
The drone replaces the *means of access*. Without closing lanes or building scaffolds, it carries a camera right up to the member. There are three main ways to use it.
| Approach | Typical aircraft | Best for |
|---|---|---|
| Photogrammetry & overview | GNSS drone with high-resolution camera | Deck, railings, pier faces, 3D models |
| Close-range under-deck imaging | Upward gimbal, vision-positioned or autonomous drones (e.g. Skydio X10 class) | GNSS-weak soffits and deck undersides |
| Collision-tolerant | Caged drones (e.g. Flyability Elios 3) | Inside box girders, tight bearing areas, darkness |
| Contact sensing | Drones with arms (hammering, ultrasonic thickness) | Sounding substitutes, steel thickness (mostly trials) |
The basic pattern is a grid: hold a constant stand-off from the surface and overlap images in both directions (60–80% overlap is common). The images can then be stitched into orthomosaics or 3D models, so cracks can be mapped onto drawings with position and length.
Overcoming the environment
Under a bridge is a hostile place for drones. The deck blocks GNSS, so position drifts; steel girders disturb the magnetic field, so the compass lies. Over rivers wind funnels through the valley, and water reflections can fool downward range sensors.
- GNSS-denied: choose aircraft that localise from cameras or LiDAR (visual-inertial odometry, SLAM). GNSS-denied environments
- Magnetic disturbance: check compass health before take-off and be ready to fly in modes that do not rely on magnetic heading. Magnetic disturbance
- Wind: air wraps around piers. Keep an escape path away from the structure and extra stand-off margin. Wind & turbulence
- Darkness: box girders and girder ends are pitch black. Use onboard lighting and fixed exposure to keep image quality consistent. Darkness
From images to decisions
Taking pictures does not finish an inspection. From thousands of images you must find defects, measure them, locate them, compare with last time, and rate the bridge. AI helps with finding, measuring and comparing; rating remains the responsible engineer’s job.
| Step | What machines do | What people own |
|---|---|---|
| Detection | Propose cracks, spalls, rust from images (deep learning) | Check misses and false alarms |
| Measurement | Convert pixels to crack width/length, rust area | Confirm scale and capture conditions |
| Localisation | Place defects on 3D models or damage maps | Match to member IDs |
| Change detection | Differences from the last inspection | Interpret what the progression means |
| Rating | — | Condition rating, actions, signing the record |
Inspection data becomes valuable only when it accumulates per bridge, per member, per year. If every flight produces a different format, nothing can be compared. Tying member ID, camera pose and capture conditions to every image from day one is the best investment in future inspection.
Rules and responsibility
Japan’s periodic inspection guidelines were revised in February 2019 so that, while close visual inspection remains the basis, inspection-support technologies such as drones can be used to inspect more efficiently. The guideline is generally described as allowing methods the inspector judges to give a diagnosis equivalent to close visual inspection. MLIT publishes a performance catalogue of inspection-support technologies so owners can compare them (the 2019 draft started with 16 technologies).
Flying is regulated separately. In Japan, flights over densely inhabited districts or within 30 m of people or property are “specified flights” that require permission or approval depending on conditions, plus a flight-plan notification and a flight log. Bridges cross roads, railways and rivers, so coordination with road, river and rail authorities is essential. Compare countries in world regulations.
What drones cannot do
- Hammer sounding & touch: delamination is traditionally found by tapping. Drone-mounted sounding devices are being tested but are not yet a general substitute.
- Corrosion under coatings, internal flaws: photos show surfaces only. Thickness and internal cracks need ultrasonic, eddy-current or other NDT.
- Areas hidden by mud or nests: clean first. What you cannot see, you cannot photograph.
- Rating and accountability: interpreting defects and deciding actions remains a human job.
2035: the world where inspection works
At the foot of a town bridge sits a small drone port. Before dawn, a drone threads the girders on a fixed route and photographs each member from the same pose under the same conditions as last time. By noon, only the members that changed appear on the engineer’s screen. The five-yearly “big inspection” has become yearly watching.
Nobody needs to climb the bridge. People interpret the changes, decide the order of repairs and explain it to residents. Inspection shifts from a job of going to dangerous places to a job of caring for the whole life of the bridge.
PREPARATION
Pre-flight preparation checklist
Ticks are saved on this device (and to your shared account if signed in). This is a learning aid, not proof of permission or safety.
Cases for this asset
- MnDOT phased UAS bridge inspection research on 39 bridges, including the Blatnik BridgeDuluth, Minnesota, United States (plus Nielsville, St. Paul and Stillwater) · 2016
- MLIT Kinki Regional Bureau: cost and schedule comparison for bridge inspection with a 360° obstacle-sensing droneKinki region, Japan (Aji Viaduct 750 m, Yamatogawa Bridge 250.5 m, Yamatogawa-kita Viaduct 510.5 m) · 2025
Related papers
- Deep Learning-Based Crack Damage Detection Using Convolutional Neural NetworksCha, Y.-J. 2017
- Autonomous Structural Visual Inspection Using Region-Based Deep Learning for Detecting Multiple Damage TypesCha, Y.-J. 2018
- Autonomous UAVs for Structural Health Monitoring Using Deep Learning and an Ultrasonic Beacon System with Geo-TaggingKang, D. 2018
- Advances in Computer Vision-Based Civil Infrastructure Inspection and MonitoringSpencer, B. F. 2019
- A review on computer vision based defect detection and condition assessment of concrete and asphalt civil infrastructureKoch, C. 2015
- Visual monitoring of civil infrastructure systems via camera-equipped Unmanned Aerial Vehicles (UAVs): a review of related worksHam, Y. 2016
- Bridge inspection: human performance, unmanned aerial systems and automationDorafshan, S. 2018
- UAV-Based Remote Sensing Applications for Bridge Condition AssessmentFeroz, S. 2021
- Unmanned Aerial Vehicles (UAVs): A Survey on Civil Applications and Key Research ChallengesShakhatreh, H. 2019
- Aerial Manipulation: A Literature ReviewRuggiero, F. 2018
- Past, Present, and Future of Aerial Robotic ManipulatorsOllero, A. 2022
- Application of Crack Identification Techniques for an Aging Concrete Bridge Inspection Using an Unmanned Aerial VehicleKim, I.-H. 2018
- Quality Evaluation of Digital Twins Generated Based on UAV Photogrammetry and TLS: Bridge Case StudyMohammadi, M. 2021
- Image-based crack assessment of bridge piers using unmanned aerial vehicles and three-dimensional scene reconstructionLiu, Y.-F. 2020
- Vision and Control for UAVs: A Survey of General Methods and of Inexpensive Platforms for Infrastructure InspectionMáthé, K. 2015
Country sources(139)
- Guidebook for Promoting Drone Use and Development in the Construction SectorJP · 国土交通省(大臣官房技術調査課系)
- Drones trialled for bridge and tunnel inspection as MLIT tests maintenance robotsJP · ニュースイッチ(日刊工業新聞社)
- NEXCO introduces steel bridge inspection using drone videoJP · NEXCO東日本・中日本・西日本
- JR Central's efforts toward introducing dronesJP · 東海旅客鉄道(名古屋産業振興公社のセミナー資料)
- MLIT expands the Inspection Support Technology Performance CatalogJP · 国土交通省 道路局
- Robot camera demonstration for bridge inspection in Misaki TownJP · 国土交通省 近畿地方整備局 大阪国道事務所・岬町
- Infrastructure inspection technology using Skydio X10 listed in MLIT's inspection technology catalogJP · PR TIMES(企業プレスリリース)
- Road management: drones to raise safety and efficiencyKR · 데일리안
- MOLIT to use drones for railway facility inspectionKR · 산업일보
- Bridge undersides to be inspected by droneKR · 전자신문
- 'Eagle-eyed' autonomous drones to inspect subway facilitiesKR · 서울특별시 내 손안에 서울
- Capturing minute cracks with drones: POS-VISION apartment wall managementKR · POSCO Newsroom
Check the evidence
Rules, figures and cases cite the sources below. Future scenarios and design proposals are the editors’ own view.
- MLIT press release: Call for new inspection technologies for bridges (30 July 2019)MLIT Road Bureau · Full text read
- MLIT: FY2024 inspection results for bridges and other structures (Road Maintenance Annual Report)MLIT Road Bureau · Full text read
- MLIT: FY2015 inspection results for bridges and other structuresMLIT Road Bureau · Full text read
- MLIT: Performance catalogue of inspection-support technologiesMLIT · Full text read
- MLIT: Permission and approval procedures for unmanned aircraft (specified flights)MLIT Civil Aviation Bureau · Full text read
- eCFR: 23 CFR Part 650 Subpart C – National Bridge Inspection StandardsU.S. Government Publishing Office / FHWA · Full text read
- FHWA: National Bridge Inspection StandardsU.S. Federal Highway Administration · Full text read
Sources checked: 2026-10-22