Project Details

Abstract

Nationwide, more than 300,000 bridges are annually inspected. In many states, both a National Bridge Inventory (NBI) inspection and an element level inspection (following the AASHTO Manual for Bridge Element Inspection, MBEI) must be completed for each bridge. Using either NBI or MBEI, a significant amount of data is collected and reported. However, the data collection and reporting are usually done manually, which are time consuming, error prone, and sometimes not consistent when repeated. Computer vision can significantly expedite damage identification and quantification using images of bridge elements. The main goal of the present study is to develop practical AI tools that help inspectors with measurements and reporting of bridge defects following NBI and MBEI requirements. To achieve this goal, a few bridge elements (e.g., decks and girders) will be targeted, inspection database including photographs of the selected elements will be compiled, and computer vision tools will be developed to detect the element defects, quantify the defect per NBI/MBEI, and produce a report following standard practices. The tools, which can be standard software or web-based, will incorporate drones and mobile devices for the ease of data collection, access, sharing, and reuse in future inspections.

Project Word Files