A remote bridge-assessment framework combines UAV imagery, Structure from Motion point clouds and RANSAC-based region extraction, comparing M3C2, C2C and C2M for translation, rotation and settlement. Experiments report high accuracy and efficiency, but the abstract provides no error values or real post-disaster deployment evidence.
Key findings
- The framework measured simulated translation, rotation and settlement and identified which point-cloud comparison method was more suitable for each movement type. High accuracy and efficiency were claimed, without numerical errors or processing times in the abstract.
Why this matters globally
Post-disaster bridge assessment is a global recovery bottleneck. An accessible UAV-software approach could reduce time and exposure, especially where permanent sensors are scarce.
Thai researcher contribution
Chulalongkorn and Kasetsart University researchers integrated infrastructure engineering, remote sensing and machine learning.
Limitations to consider
The abstract omits movement magnitudes, ground truth, repetitions and lighting, wind, surface and occlusion conditions. Point-cloud registration is critical, and controlled experiments are not real earthquake deployment. The method cannot replace structural-engineer inspection.