Multimodal bridge inspection using drone, GPR, and infrared thermography for safe, fast, and smart infrastructure assessment

By Dr Sherif Abdelkhalek, Dr Nour Faris and Prof Tarek Zayed

Bridges form the backbone of contemporary transport infrastructure, connecting communities, enabling commercial activities, and supporting everyday mobility. However, these critical assets deteriorate due to continuous exposure to harsh environmental conditions, heavy traffic loads, pollution, and accidental impacts 1. This progressive deterioration manifests as concrete cracking, delamination, reinforcement corrosion, and structural fatigue. If left unaddressed, these deficiencies may jeopardise safety, reduce service life, and ultimately result in partial or catastrophic failure, thereby causing traffic disruption, economic losses, and, in the most severe circumstances, loss of life.

Concrete bridge defects

Hong Kong clearly illustrates this challenge. The city has approximately 1,340 vehicular bridges, of which around 7% are more than 50 years old and about 40% are over 30 years old. Concrete bridges are the most common bridge type in Hong Kong, accounting for approximately 88% of the existing bridges. The high temperatures, extreme level of humidity, and heavy rainfall in Hong Kong accelerate the deterioration of concrete bridges, causing cracking, corrosion, and spalling¹. The dense urban environment further complicates the situation: access is restricted, lane closures are disruptive, and arranging scaffolding or under‑bridge inspection units is often difficult and costly. The socio‑economic risks became particularly evident after a barge struck the Kap Shui Mun Bridge in October 2015, severely disrupting the main transport link to Hong Kong International Airport and raising urgent concerns about structural integrity and network resilience.

Given these risks, bridge authorities typically conduct regular condition assessments every one to two years. The objective is to identify emerging defects early, schedule timely interventions, and allocate maintenance budgets efficiently. Despite the importance of this procedure, traditional methods primarily rely on visual inspection². In this approach, various bridge components are examined visually by an experienced inspector. The inspector usually uses basic tools for defect investigation, such as a measuring tape, a marker, chalk, and flashlight.

Age distribution of footbridges and vehicle bridges in Hong Kong

Why visual inspection is no longer enough

Visual inspection remains the backbone of most bridge assessment protocols (for example, ASTM and the U.S. National Bridge Inspection Standards). Rightly so, it effectively identifies surface-level defects such as cracks, exposed rebar, efflorescence, or spalled concrete. Nonetheless, visual inspection has critical limitations: 1) human inspectors, no matter how skilled, cannot see beneath the surface, for example, subsurface corrosion, voids, or delamination may remain undetected until they manifest as visible distress, often too late for cost-effective intervention; 2) visual inspections are inherently subjective, and may vary depending on the inspector’s experience, lighting conditions, and accessibility constraints; and 3) in complex urban environments such as Hong Kong, where erecting scaffolding or implementing lane closures is costly and disruptive, traditional methods become impractical and potentially hazardous. For engineers tasked with prioritising repairs under tight budget constraints, this lack of early, objective, and subsurface data often leads to suboptimal decision-making, for example, either overspending on unnecessary interventions or underestimating risk until major repair (or failure) occurs.

The Need for Advanced Inspection Technologies

The engineering community urgently needs a smarter, faster, and more reliable approach that combines advanced sensing technologies with automation and intelligent data interpretation. At The Hong Kong Polytechnic University (PolyU), a research team led by Prof Tarek Zayed (from the Department of Building and Real Estate) has developed a ground-breaking intelligent bridge inspection system. This innovative solution efficiently detects surface cracks and uncovers subsurface structural defects that are invisible to the naked eye. The system advances conventional practices through the integration of three cutting-edge technologies: drone, Ground Penetrating Radar (GPR), and Infrared Thermography (IRT). This integrated inspection system provides a safer, faster, and more objective assessment of concrete bridges and similar structures.

The power of three

The inspection system developed by the PolyU overcomes the limitations of traditional approaches by deploying these three complementary technologies, each addressing different aspects of bridge deterioration.

1. Drone: The eye in the sky

Drones, equipped with high-resolution RGB cameras, serve as the first line of inspection. They provide rapid, safe, and detailed visual documentation of the entire bridge structure, including hard-to-reach areas such as undersides of decks, tall piers, and expansion joints.

Modern engineering-grade drones offer:

1. Centimetre-level positioning accuracy using RTK/PPK GPS.
2. Automated flight planning for repeatable, and systematic coverage.
3. 4K or higher-resolution imagery, enabling accurate crack width measurement.
4. Thermal camera integration, allowing dual RGB and thermal data collection in a single flight.

For engineers, drones significantly reduce inspection time, eliminate risks associated with working at heights, and minimise traffic disruptions. A bridge that once required days of lane closures and scaffolding can now be visually surveyed in a few hours.

2. Ground-penetrating radar: Seeing beneath the surface
While drones capture the surface conditions, GPR reveals what lies beneath. This non-destructive testing (NDT) method uses high-frequency electromagnetic pulses (typically 900 MHz to 2.6 GHz for concrete applications) transmitted into the structure via an antenna. A typical GPR system used for bridge deck inspection consists of three main components:

1. an antenna, which transmits and receives electromagnetic waves;
2. a processing and display unit, which records the signals and converts them into interpretable images; and
3. a distance‑measuring device (such as an odometer wheel or encoder) that tracks the antenna’s position along the scan path. These components are typically mounted on a cart or autonomous robot to facilitate efficient bridge deck inspections.

GPR scanning using (a) cart setup and (b) handheld setup

The operational principle of GPR relies on the sensitivity of electromagnetic waves to the dielectric properties of different materials 3. When the antenna is pushed or dragged across the concrete surface, it emits electromagnetic waves that penetrate the concrete. When these waves encounter interfaces between materials with different properties, such as transitioning from air to concrete, or from concrete to reinforced rebar, they reflect back to the antenna. GPR data provide two critical pieces of information 4: Two-Way Travel Time (TWTT), which indicates how long it takes for the waves to travel through the concrete and return; and amplitude, which reflects the strength of the returned signal. By analysing these two parameters, GPR can identify hidden conditions within the concrete, including areas affected by moisture ingress and corrosion, even before visual symptoms appear. For instance, corrosion creates a conductive environment that damps the radar signal, resulting in reduced amplitude and distorted reflection patterns from the rebar. By analysing these changes, engineers can identify and map corrosionprone zones, enabling targeted interventions. This capability is invaluable for the early detection of potential structural issues. However, interpretation of GPR signals requires specialised expertise, highlighting the need for automated analysis tools, which will be discussed later.

3. Infrared thermography: Mapping thermal anomalies

Infrared Thermography (IRT) is another non‑destructive technique that detects surface temperature variations to infer subsurface conditions⁵. The method is based on three fundamental physical principles: 1) any object with a temperature above absolute zero emits thermal radiation; 2) the intensity of this radiation depends on the object’s temperature and emissivity; and 3) defects, such as voids and delamination, disrupt heat flow within the material, causing localised temperature differences at the surface.

In concrete bridge decks, delaminated regions or near‑surface voids tend to heat up and cool down at different rates compared with sound concrete 6. When the deck is exposed to solar heating during the day, delaminated zones often appear as “hot spots” in thermal images; after sunset, they cool more rapidly and may appear as “cold spots”. IRT cameras capture these temperature patterns over large areas in a very short time. When used correctly, typically under suitable environmental conditions and with appropriate calibration, IRT can reveal defects such as delamination, debonded overlays, near‑surface voids, scaling, and zones of severe cracking.

Integration of complementary technologies for holistic concrete bridges assessment

Individually, these inspection technologies are valuable, however, their true power emerges when they are integrated into a unified inspection workflow. Each of the technologies brings unique strengths, enabling engineers to evaluate bridges more effectively. For example, drone cameras can quickly assess surface issues, such as cracks or spalling, providing immediate visual feedback on the condition of the bridge surface. GPR extends this assessment goes a step further by mapping hidden corrosion beneath the surface, which is typically not visible in digital imagery. IRT complements these methods by identifying areas of delamination, revealing potential structural issues that could compromise safety. By utilising GPR and IRT alongside visual inspections (using a drone camera), the team was able to collect critical data that provide deeper insight into the bridge's condition. This comprehensive approach enables conclusions to be drawn based on objective evidence captured from both the surface and the interior of the concrete, while achieving greater time efficiency.

Integrated inspection system

Streamlined data collection and analysis

The fundamental breakthrough in the proposed inspection system occurs in both the data collection and analysis stages. With respect to data collection, drone was used to automate the visual surveys using digital and thermal cameras, while a robotic platform (ground robot) was developed to automate data gathering with GPR. Combining these technologies enabled automated and efficient data collection, resulting in more consistent and effective inspections.

Automate data collection using (a) a drone with thermal camera and (b) GPR system mounted on an autonomous ground robot

However, collecting terabytes of data is meaningless without actionable insight. The PolyU team recognised that the most significant barrier to adopting these technologies is not hardware, but rather the complexity of data interpretation 2,7. Therefore, on the analytical side, the research team developed a unified software tool that capable of importing data from GPR, IRT, and drone imagery. The tool employs advanced signal-processing routines and AI-based algorithms to clean and filter raw signals, extract key features associated with cracks, corrosion, and delamination, and automatically classify the condition of various zones. This approach greatly reduces the need for manual interpretation and improves repeatability across inspections and between different inspectors. The user-friendly tool enhances accuracy and overcomes the core challenges associated with applying such technologies in bridge inspection, while concealing the complexity of the underlying algorithms behind intuitive dashboards and straightforward outputs.

Corrosion analysis section in the developed software

From data to decision: Condition maps and health indices

For bridge authorities, raw data are only useful if they can be translated into clear maintenance decisions. To support this, the PolyU team presents processed data in the form of colour‑coded corrosion and delamination maps. The corrosion map derived from GPR data categorises corrosion severity into four levels based on the properties of the reflected waves from rebar: sound (no colour), mild corrosion (green), moderate corrosion (yellow), and severe corrosion (red). Similarly, the delamination map, which is generated from thermal imaging data, illustrates three severity levels: sound (no colour), moderate (yellow), and severe (red).

Condition mapping obtained from (a) GPR scanning and (b) IR scanning

By overlaying these maps on plan views of the concrete surface, such as a bridge deck, engineers can quickly identify areas where subsurface defects are concentrated. The interrelations between detected corrosion and delamination zones can be easily examined, since reinforcement corrosion is often accompanied by near‑surface delamination in the same zones. This cross‑validation increases confidence in the diagnostic accuracy. These maps are not just visual aids; they form integral components of Bridge Management Systems (BMS). Utilising these datasets allows engineers to track deterioration over time, quantify required repairs, prioritise work orders based on risk assessments, and generate comprehensive reports for stakeholders.

To further aid decision‑making, the research team has also developed condition rating scales and an overall health index that combine information from surface defects, corrosion, and delamination. These indices translate spatial condition data into simple numerical scores (for example, 0–100 scale), enabling consistent condition rating across assets and over time. This empowers asset managers to prioritise repairs across a network of structures and justify funding decisions using objective data rather than purely subjective assessments.

Condition indices for various defects and overall

Scalability and applications beyond bridges

Although the system was initially developed for concrete bridge decks in Hong Kong, its architecture is highly scalable. The system is capable of supporting a wide range of inspection tasks, from small footbridges to large vehicular bridges. The integration of a camera mounted on a drone significantly enhances the system's versatility, enabling access to hard-to-reach areas, such as long bridge columns and the underside of bridge decks, that would typically present challenges for traditional inspection methods. This capability is particularly valuable in urban environments such as Hong Kong, where infrastructure is often extensive and complex, making manual inspections more difficult and timeconsuming. By utilising drone technology, the inspection system not only speeds up the evaluation process but also mitigates safety risks associated with working at height or in restricted locations.

Moreover, the robustness of the proposed system allows it to be customised to meet specific inspection requirements across different structures. For instance, the system can be deployed across various structural elements and types of infrastructure, including bridges, tunnels, and large buildings, making it adaptable to diverse engineering contexts. To further enhance scalability, the user‑friendly software encapsulates complex processing steps within standardised workflows, significantly lowering the barrier to adoption for government agencies, consultants, and contractors. This design enables engineers to focus on interpreting results and planning interventions, rather than grappling with raw data.

Conclusion

Aging bridges require more than routine inspections; they demand diagnostic accuracy. The incorporation of drone, GPR, and IRT provides engineers with an unparalleled capacity to detect hidden defects, quantify uncertainties, and implement preventative measures prior to failure. This represents not merely a technological enhancement but a fundamental shift towards proactive, predictive, and performance-oriented asset management. Like other inspection techniques, these technologies have limitations. Drones may face no-fly restrictions near airports or be constrained by high winds; while IRT requires sufficient thermal gradients, which may not be present under certain weather conditions. Despite these limitations, the combined application of these technologies offers significant potential to enhance inspection outcomes. The continued advancement of automated data interpretation and AI-driven analytics will further amplify their effectiveness, paving the way for smarter infrastructure stewardship in the years ahead.

References

1. A. Ibrahim, S. Abdelkhalek, T. Zayed, A.H. Qureshi, E.M. Abdelkader, A Comprehensive Review of the Key Deterioration Factors of Concrete Bridge Decks, Buildings 14 (2024). https://doi.org/10.3390/buildings14113425.
2. S. Abdelkhalek, T. Zayed, Comprehensive Inspection System for Concrete Bridge Deck Application: Current Situation and Future Needs, J. Perform. Constr. Facil. 34 (2020). https://doi.org/10.1061/(ASCE)CF.1943-5509.0001484.
3. N. Faris, T. Zayed, A. Fares, S. Abdelkhalek, E.M. Abdelkader, Automated rebar recognition and corrosion assessment of concrete bridge decks using ground penetrating radar, Autom. Constr. 166 (2024) 105631. https://doi.org/https://doi.org/10.1016/j.autcon.2024.105631.
4. N. Faris, A.K. Khalil, M.A.A. Abdelkareem, S. Abdelkhalek, A. Fares, T. Zayed, G. Alfalah, A GPR-based framework for assessing corrosivity of concrete structures using frequency domain approach, Heliyon 11 (2025). https://doi.org/10.1016/j.heliyon.2025.e42641.
5. A. Ibrahim, F. Nour, Z. Tarek, Q. Abdul Hannan, A. Sherif, E.M. and Abdelkader, Application of infrared thermography in concrete bridge deck inspection: current practices, challenges and future needs, Nondestruct. Test. Eval. (2026) 1–44. https://doi.org/10.1080/10589759.2024.2443810.
6. A.H. Qureshi, T. Zayed, A. Ibrahim, C. Xiong, S. Abdelkhalek, B. Manzoor, Optimum thermal gradient thresholds (OTGT) system for inspecting bridges using Infrared Thermography (IRT), Adv. Eng. Informatics 67 (2025) 103520. https://doi.org/https://doi.org/10.1016/j.aei.2025.103520.
7. S. Abdelkhalek, T. Zayed, Performance assessment model of non-destructive technologies in inspecting concrete bridge decks, Struct. Infrastruct. Eng. 19 (2023) 216–237. https://doi.org/10.1080/15732479.2021.1937234.