Application of Crack Identification Techniques for an Aging Concrete Bridge Inspection Using an Unmanned Aerial Vehicle

Bridge inspection using unmanned aerial vehicles (UAV) with high performance vision sensors has received considerable attention due to its safety and reliability. As bridges become obsolete, the number of bridges that need to be inspected increases, and they require much maintenance cost. Therefore, a bridge inspection method based on UAV with vision sensors is proposed as one of the promising strategies to maintain bridges. In this paper, a crack identification method by using a commercial UAV with a high resolution vision sensor is investigated in an aging concrete bridge. First, a point cloud-based background model is generated in the preliminary flight. Then, cracks on the structural surface are detected with the deep learning algorithm, and their thickness and length are calculated. In the deep learning method, region with convolutional neural networks (R-CNN)-based transfer learning is applied. As a result, a new network for the 384 collected crack images of 256 × 256 pixel resolution is generated from the pre-trained network. A field test is conducted to verify the proposed approach, and the experimental results proved that the UAV-based bridge inspection is effective at identifying and quantifying the cracks on the structures.