At STRUCINSPECT, we have developed an innovative, AI-based method to improve the status quo for structural inspection of civil engineeringstructures. The approach is currently focused heavily on concrete constructions, especially bridge inspection and aims to act as a supporting tech solution to help inspection teams make their work more efficient, safer and more sustainable. The technology is centered around offering multiple services and benefits, including structural inspections, digital twin creation and Building Information Modeling (BIM), AI-based damage analysis, visualization, collaboration and sharing of results as well as lifecycle analysis and asset management. STRUCINSPECT was founded in 2019 as a joint venture between PALFINGER, VCE and ANGST .
Background
Engineering structures have to fulfill stringent requirements regarding stability, traffic safety and durability. As a result, these structures are subject to periodical inspections to ensure ongoing compliance and safety. Most countries have laws and regulations in place mandating this sort of continued maintenance.
The innovative, AI-based inspection platform we built at STRUCINSPECT creates digital twins and BIM models using photogrammetry. On the back of 3D models, damage is detected, can be classified and documented more accurately and efficiently than previously possible by using Artificial Intelligence support. The platform goes further than acting as a mere processing tool. It also fulfills the function of a communication and presentation tool for all project participants. Below, we dive into the players that are making it happen and offer you a detailed look at the various benefits the solution provides.
A collaborative AI-based platform for engineers and infrastructure operators
The STRUCINSPECT joint venture
STRUCINSPECT is a joint venture between PALFINGER, VCE, and the ANGST Group, which makes for three specialists from three different fields of civil engineering. VCE works as a high-tech oriented engineering company, specialized in condition inspections and evaluations of infrastructure constructions. PALFINGER provides specialized equipment for structural inspections, enabling access of inspectors and civil engineers to hard-to-reach places. Finally, the ANGST Group are specialists in mobile mapping, professional drone operation and photogrammetry. Together, these three companies have developed the AI-supported platform for collaborative structural inspection of the future.
The STRUCINSPECT platform for collaborative, interdisciplinary work
On the STRUCINSPECT platform, engineers executing inspections can perform a complete, AI-supported digital structural analysis while simultaneously sharing intermediate as well as final results with the infrastructure operator. This enables maximum efficiency and saving time. The final result can be shared directly with clients or owners of constructions as well. In addition, building owners can share all results with third parties such as engineering firms and construction companies, and use them for further joint work, including maintenance or repair. Further value can be extracted from results by entering these into an asset management system and maintaining accurate records of inspection data.
Data acquisition
The basis for the fully detailed analyses is the full-surface recording of the entire structure with high-resolution images according to specified quality criteria. Building inspectors can acquire the images themselves or request and commission support from experts for image acquisition via the STRUCINSPECT platform. The experts that can be tapped capture the required images in the air, using Unmanned Aerial Vehicles (UAVs), on land and on water (by boat), depending on project needs and accessibility. The data collected in this way is automatically checked for completeness and quality after being uploaded to the STRUCINSPECT platform.
Data quality and completeness are essential for further processing, such as AI-supported damage analysis and the creation of an accurate digital twin and an Industry Foundation Class (IFC)-compliant BIM model. To this end, STRUCINSPECT has developed a binding quality and interface catalog with hard criteria such as resolution, overlap, Ground Sampling Distance (GSD) or focus type that must be met before the data is accepted for further processing.
Manage damage images, Digital twin and BIM
When creating the digital twin from the recorded images, the STRUCINSPECT platform enables the selection of different quality levels. The basic version allows to create structures and enables to upload, rate and manage damages in an easy way. With a 3D twin as a point cloud, which allows georeferencing for the precise localization of damage another option is available.
The premium version includes the creation of a high-resolution textured 3D twin, which is so accurate that it allows virtual manual inspection and 3D damage control. Within the premium version also the creation of an IFC-compliant BIM model at the selected Level of Detail (LOD) is possible. In addition, all required and existing structural data such as use, age of the structure, type of structure or length of a bridge is recorded for all 3D models.
AI damage analysis
The AI-based damage detection that we developed can detect cracks, spalling, exposed reinforcement and corrosion , rust stains, cavity, moisture, gravel pockets and moss with a detection rate (TPR) of 99.9%, fully automatically. In addition to the detection and marking of the damage, an automatic geometric analysis (e.g., crack width measurement, designation of the damage area) is also performed. The desired accuracy criteria and classifications can be defined in advance. For example, a defined crack width of 0.1 mm for pretensioned concrete structures and 0.3 mm for reinforced concrete structures can be specified. Cracks and hair cracks that are visually interrupted but related are recognized as one coherent damage. Using an adjustable minimum size, spalls are also automatically recognized, marked and measured as damage by the AI. This all is the base for the further assessment of an expert performing the inspection.
AI-based assignment, documentation and visualization service
With the help of AI, the detected damage on 2D images (original images) and 3D images (georeferenced digital twin) is assigned and documented with its own unique identification number. If the same damage is also detected on other images and from other angles, automatic matching and automatic assignment to the identification number takes place. In other words, the same damage, even if it is detected on several images, is recognized as unique and only entered once in the damage database. All damages, along with their unique identification number, are also georeferenced on the 3D digital twin.
Findings toolbox, results sharing and asset management
All discovered damages are stored in the dedicated database and can be edited in 2D and 3D with just a few clicks. An inspection engineer can accept, modify, extend or reject damage recommendations and assessments made via the platform. Possible follow-up measures can also be added as action items.
Further, the structural inspectors can provide defined access to the generated results to their customers, the infrastructure operators, and – upon request of their customers – also to other, third party service partners. This helps to ensure the highest level of transparency and simplifies the collaboration between all involved parties. The findings, which can be viewed by the infrastructure operators, include damage catalogs, ratings, recommended measures, and more. All results can also be downloaded as a damage dataset, BIM dataset or PDF report.
Direct or indirect connection to the customer’s existing building databases and asset management systems is also possible to enable ease of data exchange. In addition, the STRUCINSPECT platform enables the execution of life cycle analyses and maintenance planning based on various aging and cost models. A management center enables every user to have a simple overview of all their projects, structures and data.
Added Value
Compared to conventional state-of-the-art structural inspection, the STRUCINSPECT platform is a time-saving, effective and efficient evolution and base for the enormous need of digitalization in the inspection business. The main advantages of STRUCINSPECT include the degree of objectivity and repeatability of the end-to-end processing involved and the comprehensive nature of the documentation of the condition of the entire surface of a structure (even damage-free areas are fully documented). Finally, the platform offers the significant benefit of simple, automated documentation tracking of the damage inspection progress during periodic, follow-on inspections.
Outlook and Further Developments
The STRUCINSPECT platform is constantly being developed further, allowing us to respond to explicit requests, feedback and input from users in a continuous and timely manner. Other planned features in the current development pipeline include additional asset management features with collaborations or plug-ins for third-party asset management software systems, the inclusion and testing of additional asset classes (such as roads and tunnels), and the integration of a more in-depth, multi-spectral data collection and evaluation process to be able to include and assess additional damage types.