Damage Detection of Bridges Using Deep Learning and Unmanned Aerial Systems

Abstract: Bridges are an essential part of the world's infrastructure, with millions built over millennia. But a lot of these bridges are getting older and come with a lot of concerns. Regularly, risk-based inspections and maintenance are essential to ensuring public safety, but existing management techniques mainly rely on manual processes, which create inefficiencies. This antiquated method raises the risk of bridge deterioration, which lowers the bridges' capacity to support loads and jeopardizes their functionality and safety.
Structural Health Monitoring (SHM) offers an advanced approach to bridge inspections, enhancing serviceability, safety, and extending the lifespan of bridges. SHM employs various techniques to assess structural health, aiding stakeholders in making informed decisions to achieve resilience. Deep Learning (DL) has recently been integrated into SHM, providing powerful tools for estimating traffic flow and evaluating bridge conditions.
Traditional SHM methods, while effective, can be further optimized through modern technologies. One such innovation is the use of Unmanned Aerial Systems (UAS), or drones, which allow for detailed bridge inspections without disrupting traffic. UAS can access hard-to-reach areas more efficiently than conventional methods, enabling comprehensive assessments. This research focuses on applying SHM to an existing bridge inspected by UAS, with a state-of-the-art DL model used to monitor its condition. The results are compared to those from other models to demonstrate the enhanced performance of this approach.