Traffic Disruption-free Bridge Inspection Initiative with Robotic Systems

The current practice of bridge inspection requires the use of snooper trucks and, when parked on bridges, the control of passing traffic, thus causing an operation safety concern for both passengers and inspectors. It also leads to inherently subjective and qualitative results. This pooled-fund study aims to develop case studies on the deployment and performance of Bridge Inspection Robot Deployment Systems (BIRDS) developed in the INSPIRE University Transportation Center for faster, safer, cheaper, and more quantitative bridge inspection with minimum impact on traffic flow. To this endeavor, an automated bridge preservation framework was envisioned to integrate advanced robotics, remote sensing, and nondestructive testing into the practice of visual inspection and associated maintenance. By evaluating the advanced technologies at 59 bridges in diverse types, age groups, and geographical locations, the best practices of the technologies were summarized in inspection protocols and guidelines using commercial drones, structural crawlers, and custom-built hybrid uncrewed vehicles. Vision-based instance segmentation via machine learning efficiently and effectively detected, located, and quantified weld defects in steel bridges, including cracks, debonding, and cavity, in real time. Topside and underside deck inspections were compared to ensure the reliability of traffic disruption-free bridge inspection from the underside of the bridge deck. By combining flying, traversing, and crawling capabilities, the award-winning invention - BIRDS offered a versatile robotic solution that addressed the limitations of commercial drone technologies. Its ability to seamlessly transition between aerial and ground-based inspection modes ensured a comprehensive coverage of bridge structures. This innovation enabled both global visual monitoring and local detailed inspection using remote sensing (e.g., microscope imager and laser scanner) and nondestructive testing (e.g., ultrasonic metal thickness gauge) for the detection and quantification of bridge surface and substrate defects. Since a limited type and number of defects were observed from the selected bridges, more bridges should be inspected to collect big data required in machine learning to develop decision-making support tools toward data-driven bridge asset management.

VIEW FINAL REPORT

Report number: cmr 25-011
Published: September 2025

Project number: TR202004

Authors: Genda Chen, Zhenhua Shi, Son Nguyen, Mohammad Hossein Afsharmovahed, Peter Damilola Ogunjinmi, Ying Zhuo

Performing organization: Missouri University of Science & Technology