Timely assessment of integrity of structures after seismic events is crucial for public safety and emergency response. This study focuses on assessing the structural damage conditions using deep learning methods to detect exposed steel reinforcement in concrete buildings and bridges after large earthquakes. Steel bars are typically exposed after concrete spalling or large flexural or shear cracks. The amount and distribution of exposed steel reinforcement is an indication of structural damage and degradation. To automatically detect exposed steel bars, new datasets of images collected after the 2023 Turkey Earthquakes were labeled to represent a wide variety of damaged concrete structures. The proposed method builds upon a deep learning framework, enhanced with fine-tuning, data augmentation, and testing on public datasets. An automated classification framework is developed that can be used to identify inside/outside buildings and structural components. Then, a YOLOv11 (You Only Look Once) model is trained to detect cracking and spalling damage and exposed bars. Another YOLO model is finetuned to distinguish different categories of structural damage levels. All these trained models are used to create a hybrid framework to automatically and reliably determine the damage levels from input images. This research demonstrates that rapid and automated damage detection following disasters is achievable across diverse damage contexts by utilizing image data collection, annotation, and deep learning approaches.

Structural integrity is vital for maintaining the safety and longevity of concrete infrastructures such as bridges, tunnels, and walls. Traditional methods for detecting damages like cracks and spalls are labor-intensive, time-consuming, and prone to human error. To address these challenges, this study explores advanced data-driven techniques using deep learning for automated damage detection and analysis. Two state-of-the-art instance segmentation models, YOLO-v7 instance segmentation and Mask R-CNN, were evaluated using a dataset comprising 400 images, augmented to 10,995 images through geometric and color-based transformations to enhance robustness. The models were trained and validated using a dataset split into 90% training set, validation and test set 10%. Performance metrics such as precision, recall, mean average precision (mAP@0.5), and frames per second (FPS) were used for evaluation. YOLO-v7 achieved a superior mAP@0.5 of 96.1% and processed 40 FPS, outperforming Mask R-CNN, which achieved a mAP@0.5 of 92.1% with a slower processing speed of 18 FPS. The findings recommend YOLO-v7 instance segmentation model for real-time, high-speed structural health monitoring, while Mask R-CNN is better suited for detailed offline assessments. This study demonstrates the potential of deep learning to revolutionize infrastructure maintenance, offering a scalable and efficient solution for automated damage detection.

Impact-sounding (IS) and impact-echo (IE) are well-developed non-destructive evaluation (NDE) methods that are widely used for inspections of concrete structures to ensure the safety and sustainability. However, it is a tedious work to collect IS and IE data along grid lines covering a large target area for characterization of subsurface defects. On the other hand, data processing is very complicated that requires domain experts to interpret the results. To address the above problems, we present a novel robotic inspection system named as Impact-Rover to automate the data collection process and introduce data analytics software to visualize the inspection result allowing regular non-professional people to understand. The system consists of three modules: 1) a robotic platform with vertical mobility to collect IS and IE data in hard-to-reach locations, 2) vision-based positioning module that fuses the RGB-D camera, IMU and wheel encoder to estimate the 6-DOF pose of the robot, 3) a data analytics software module for processing the IS data to generate defect maps. The Impact-Rover hosts both IE and IS devices on a sliding mechanism and can perform move-stop-sample operations to collect multiple IS and IE data at adjustable spacing. The robot takes samples much faster than the manual data collection method because it automatically takes the multiple measurements along a straight line and records the locations. This paper focuses on reporting experimental results on IS. We calculate features and use unsupervised learning methods for analyzing the data. By combining the pose generated by our vision-based localization module and the position of the head of the sliding mechanism we can generate maps of possible defects. The results on concrete slabs demonstrate that our impact-sounding system can effectively reveal shallow defects.

In the U.S. alone, the construction industry creates around $1.3 trillion worth of structures every year, and employs around 7 million people. We spend much of our lives surrounded by the fruits of this labor, usually without thinking about what it takes to produce, and without a real awareness of how much we rely on those structures to be safe. We rarely enter buildings worrying that they are going to collapse, for example, or drive across a bridge fearful that it will crumble beneath our tires. That safety and public trust is important to maintain, and as the numbers of structures increase every year, that means regular inspection of more and more structures every year. A large number of the structures build rely upon concrete either in part or for the vast majority of their construction.

We used Adam as the optimizer and train the model for 6 epochs. We use transfer learning to then train the model on the training data set while measuring loss and accuracy on the validation set. As shown by the loss and accuracy numbers below, the model trains very quickly. After the 1st epoch, train accuracy is 87% and validation accuracy is 97%!. This is the power of transfer learning. Our final model has a validation accuracy of 98.4%.

In modern building infrastructures, the chance to devise adaptive and unsupervised data-driven health monitoring systems is gaining in popularity due to the large availability of data from low-cost sensors with internetworking capabilities. In particular, deep learning provides the tools for processing and analyzing this unprecedented amount of data efficiently. The main purpose of this paper is to combine the recent advances of Deep Learning (DL) and statistical analysis on structural health monitoring (SHM) to develop an accurate classification tool able to discriminate among different acoustic emission events (cracks) by means of the identification of tensile, shear and mixed modes. The applications of DL in SHM systems is described by using the concept of Bidirectional Long Short Term Memory. We investigated on effective event descriptors to capture the unique characteristics from the different types of modes. Among them, Spectral Kurtosis and Spectral L2/L1 Norm exhibit distinctive behavior and effectively contributed to the learning process. This classification will contribute to unambiguously detect incipient damages, which is advantageous to realize predictive maintenance. Tests on experimental results confirm that this method achieves accurate classification (92%) capabilities of crack events and can impact on the design of future SHM technologies.

More than 100 years ago acoustic mirrors along the coast of England were used to detect the sound of approaching German zeppelins. The concave concrete structures were designed to pick up sound waves from enemy aircraft, making it possible to predict their flight trajectory, giving enough time for ground forces to be alerted to defend the towns and cities of Britain. Invented by Dr. William Sansome Tucke and known as sound mirrors, their development continued until the mid-1930s, when radar made them obsolete. Joe Pettet-Smith set out to photograph all the remaining structures following a conversation with his father, who told him about these large concrete structures dotted along the coastline between Brighton and Dover. "When I was a child my father told me stories about my grandfather and his involvement in radar," says Pettet-Smith.

This method of concurrent 3D-printing, known as swarm printing, paves the way for a team of mobile robots to print even bigger structures in future. Developed by Assistant Professor Pham Quang Cuong and his team at NTU's Singapore Centre for 3D Printing, this new multi-robot technology was published in Automation in Construction, a top tier journal for civil engineering. The NTU scientist was also behind the Ikea Bot earlier this year where two robots assembled an Ikea chair in 8 min 55s. Using a specially formulated cement mix suitable for 3-D printing, this new development will allow for unique concrete designs currently not possible with conventional casting. Structures can also be produced on demand and in a much shorter period.

The elevator doors snap shut behind Otto Solis and his fellow ironworkers. With a quick shudder, gears kick in for a rattling 90-second ascent through the concrete structure rising at the corner of Wilshire Boulevard and Figueroa Street in downtown Los Angeles. The men huddle in the confined space. Wearing hard hats, bandannas, kneepads and gloves, they look like gladiators ready to fight. Foreman Solis and his crew of 10 belong to a class of ironworkers known as rod busters.