Abstract
Monitoring aging rock slopes near urban areas requires efficient workflows to ensure safety and reduce geotechnical risks. Conventional methods for processing topographical data involve multiple manual steps, which are time-consuming and prone to errors. Automating the entire workflow from data acquisition to classification improves efficiency and ensures more objective results. A study was conducted at the St-Michel (Francon) Quarry in Montreal, Canada, to develop and test a fully automated pipeline for monitoring rock slopes.
Over 18 months, nine surveys were conducted using LiDAR and photogrammetry to produce high-resolution point clouds. Subsampling methods were applied to reduce data density while retaining essential features. Vegetation filtering was performed using supervised machine learning algorithms, improving accuracy in identifying rock surfaces. Failure modes were recognized and classified using K-means clustering and the CANUPO algorithm. This workflow enabled the differentiation of natural failure mechanisms, such as freeze-thaw cycles, from anthropogenic effects like snow dumping.
The results demonstrate the potential of automated approaches to transform geotechnical monitoring processes. Streamlined data processing steps reduce manual effort and improve consistency in analyzing large datasets. The pipeline developed offers a scalable framework for monitoring aging rock slopes in urban settings, with potential applications in real-time hazard assessments and infrastructure risk management.