The collapse of a bridge is a sudden event that happens from time to time and it has a huge impact on our society. For many reasons, poor maintenance is the main cause of many collapsed bridges. In addition, there are cases of bridge aging. To prevent such unnecessary events, continuous bridge monitoring and timely maintenance must be performed.
 
In bridge inspection, visual inspection is the most widely used method, but it relies heavily on the experience of inspectors. It is also time-consuming, labor-intensive, costly, disruptive and even unsafe for the inspectors. To address its limitations, in recent years, there has been an increasing focus on the use of drones, which are expected to make the inspection process safer, faster, and more cost-effective. In addition, it can cover areas that are difficult for inspectors to reach. However, this approach is still in its original stage, and there are still many issues that need to be addressed before it can be truly implemented.
 
Taking the United States as an example, data from 2016 shows that the number of bridges with structural defects and problems accounts for 10% of the total number of existing bridges, and the average service life of these bridges is 43 years. For another example, in Japan, half of the bridges will have a service life of 50 years by 2030. The aging problem of bridge infrastructure is inevitable, so measures and measures need to be deployed in advance.
 
Before deploying countermeasures and measures, you need to have an accurate understanding of the current state of the bridge. This requires effective data obtained through monitoring and evaluation. Usually, the visual inspection is performed by humans. This method is time-consuming, high cost and unstable accuracy.
 
Shortboard and problems
 
In recent years, drones have become more and more frequently used in various fields. Using drones to detect bridges can solve the limitations of existing or traditional methods while increasing safety and speed. Through image acquisition equipment, a large number of digitized images and information can be obtained, which can make the bridge detection information more rigorous.
 
Speaking of applications, in reality, there are still many technical challenges. For example, the localization and detection of drones take a long time, and so on.
 
Summarizing the problems faced by applying drones to bridge health monitoring can be divided into four main areas.
 
First, can the drones in the area to be detected be reached? In practice, it is often uncertain. In addition, how does the drone accurately fly in the direction defined by GPS? In practice, drones generally use GPS navigation, but GPS signals are easily interfered with by external conditions, and there are no signals and unstable signals. How to obtain accurate and effective data when the drone enters the monitoring area It also needs to be carefully considered.
 
Second, how to skillfully apply the performance of drones to bridge detection. Generally, commercial drones are affected by factors such as self-weight and wind when flying, and it is difficult to control when driving. From this perspective, drones used for bridge detection need to have special characteristics.
 
Third, the acquisition of image data. Whether the image quality of the data obtained by the drone can meet the requirements is the most critical point.
 
Fourth, how to quickly get information about the damage when the bridge is damaged.
 
solution
 
First, the positioning of drones, especially under the bridge, is about the GPS-defined environment. In South Korea, relevant organizations have studied video-based SLAM methods to address this issue. This method is based on probe technology and uses a camera and RMU to estimate the position of the drone under the bridge. The test results obtained by this method improve the accuracy of GPS positioning and obtain the corresponding detection data information, so that the condition of the bridge can be seen very clearly.
 
Second, regarding the restrictions and limitations of the drone's accessibility, the current research result is to use drones with attachment capabilities for detection. It can be very tightly attached to the structure that needs to be detected. This drone can easily detect parts and structures with relatively large inclination.
 
Third, the use of UAV performance. If only conventional commercial drones cannot meet the bridge detection operation. At present, special drones are being researched, which are designed to implement customized drones for bridge detection, focusing on functions required for detection, including flight status, accuracy, and maneuverability.
 
Fourth, identification and detection of damage. It is used for the identification of different types of damage, mainly for the detection of cracks. The R-CNN method can be used to obtain the existing image of the concrete structure. The acquired image is placed in the analysis grid for further crack analysis. R-CNN can be divided into two modes: fast R-CNN and masked R-CNN. This involves addressing the third issue, assessing the quality of the image after acquiring it, and then finding the corresponding breakpoint.
 
At present, there are also monitoring and analysis cases of applying drones to bridge coating thickness and corrosion diagnosis. By acquiring data and analyzing, you can understand the coating thickness and material corrosion situation, and realize visual observation of these situations through fixed-point monitoring. However, there are some defects in the process, such as false alarms. In this case, a hybrid scanning system can be used, and a more comprehensive image can be observed. The accuracy can be increased by 40% and the false alarm rate can be reduced.
 
At present, many infrastructure security companies are doing this kind of project practice. The goals of these studies are based on the use of drones for better, more comprehensive, and more accurate bridge detection, which can save time, cost, manpower, and effectiveness. And other aspects have been comprehensively improved.
 
by：http://www.mmcuav.com