| Project Title | Transfer learning-based crowdsourced condition assessment of bridge inventories |
| University | Lehigh University |
| Principal Investigator(s) | Shamim Pakzad |
| Funding Source(s) and Amounts Provided (by each agency or organization) | Federal Funds, $52,633 Match, $52,633 |
| Total Project Cost | $105,266 |
| Start and End Dates | 06/01/2023 - 06/30/2025 |
| Brief Description of Research Project | Bridge condition assessment and remaining useful life estimation is critical for maintaining functionality and enhancing the resilience of existing highway infrastructure in the United States. Currently, there are approximately 617,000 bridges in the U.S. that require periodical condition assessment and maintenance (artba [2021]). The current practice employed for remaining useful life estimation involves measuring the number of load cycles that a structure of interest is subjected to through approaches such as rainflow counting of strain. This typically entails deploying a fixed network of strain gauges on a bridge. However, scalability is an issue for this paradigm, especially owing to high costs and efforts associated with deployment and maintenance of wiring, sensors and battery that powers the monitoring system. To overcome the existing challenges and extend this type of analysis to virtually all bridges with little operational cost, this project proposes a large-scale field implementation of a mobile sensing-based paradigm that harnesses artificial intelligence for condition assessment of an inventory of bridges. This will facilitate the use of crowdsourced data for real-time bridge health assessment at unprecedented rates, resolution and scales. |