March 27, 2025
2 Min Read
Cutting-edge generative AI Enhances Infrastructure Maintenance and Reduces Workforce Risk
The Korea Institute of Civil Engineering & Building Technology (KICT) has developed an AI-based inspection technology integrated with autonomous drones to enhance tunnel safety and maintenance, addressing workforce shortages and data scarcity.
Addressing Challenges in Tunnel Maintenance
South Korea's aging tunnels require frequent inspection, but a declining number of specialized personnel makes traditional methods unsustainable. Deep learning models need large datasets, yet real-world images of tunnel damage are scarce.
Innovative Generative AI Inspection Technology
A research team led by Dr. Shim Seungbo at KICT has created an AI system that synthesizes realistic images of tunnel damage, enabling effective training with minimal data.
- Generates 10,000 images in 24 hours to enhance model accuracy.
- Adaptive learning refines detection capabilities, reducing training costs.
- Improves defect identification for timely maintenance.
Drones: The Future of Tunnel Inspections
AI-integrated drones successfully tested in large tunnels eliminate the need for hazardous manual inspections.
- Developed with LASTMILE, the drone operates with a 20cm margin of error.
- Features a 200m-class indoor positioning sensor for precise navigation.
- Enhances safety by reducing human exposure to dangerous conditions.
Expert Insights and Future Impact
Dr. Shim Seungbo stated, “This technology proves AI can function effectively even with limited data by generating synthetic images.”
Funded by the National Research Foundation of Korea (NRF) under the KICT Research Program, this initiative advances South Korea’s smart infrastructure goals, ensuring long-term structural integrity and public safety.
