Vertical Clearance Assessment for Highway Bridges Based on Multisensor Fusion Simultaneous Localization and Mapping

Abstract

The three-dimensional (3D) reconstruction method based on computer vision technology greatly facilitates the automated performance inspection and assessment of transportation infrastructure, in which the acquisition of vertical clearance is crucial for both traffic planning and structural integrity inspection of bridges. However, manual clearance assessment conducted by traditional measurement methods is dangerous and time-consuming. In order to address these problems, a vertical clearance assessment method for highway bridges based on multisensor fusion simultaneous localization and mapping (SLAM) was proposed. A convenient and low-cost handheld device was constructed, and a simple wireless remote data control method was proposed to improve the convenience of the device operation. The SLAM algorithm was used for the point cloud reconstruction of the scene, which could view the data acquisition process simultaneously. Based on the point cloud reconstruction data, a vertical clearance assessment method was proposed to evaluate the highway bridge's vertical clearance. The proposed method was tested through a complex section of the cross-line highway bridge. The field test results showed that the constructed equipment and the SLAM algorithm could quickly complete the bridge data acquisition and 3D point cloud reconstruction, and the process took about 316 s. The proposed vertical clearance assessment algorithm could remove the noise, such as that from vehicles, and the average vertical clearance difference between the ground truth and the extracted results was 2.5%.