A new framework for automated extraction of in-situ dangerous rock blocks based on a semi-deterministic block theory

Numerous dangerous rock blocks are located on rock slopes, potentially threatening the construction and safety operation of the adjacent engineering facilities. The dangerous rock block identification has been realized for decades in rock mechanics and engineering. However, studies on the automated identification of in-situ dangerous rock blocks (ISDRBs) are limited. In this study, a novel framework is proposed for automated extraction of ISDRBs based on a new semi-deterministic block theory considering discontinuity geometric characteristics. It involves four main steps: (1) establishment of the slope point cloud model using an unmanned aerial vehicle multi-angle photography method; (2) automated identification and information extraction of discontinuities based on various algorithms; (3) identification and geometrical characterization of in-situ rock blocks (ISRBs) based on the improved block candidates searching algorithm and polyhedral modeling; (4) stability analysis of ISRBs to extract ISDRBs based on the semi-deterministic block theory. In this framework, the actual positions, geometric characteristics, and safety factors of ISDRBs can be well-reflected, providing a quantitative reference for rockfall disaster prevention and mitigation design. The framework is applied to the stability analysis of the steep rocky slope on the left bank of the Nujiang (NJ) Bridge. The analysis results indicate that, after considering discontinuity geometric characteristics, 52.6% of the ISDRBs cannot form. Ultimately, 45 ISDRBs are identified, predominantly tetrahedron in geometry, and their volumes range from 0.02 to 32.57 \(\:{\text{m}}^{\text{3}}\), with the majority being smaller than 5 \(\:{\text{m}}^{\text{3}}\). A convenient and feasible evaluation method based on ISDRB information is finally proposed to further discuss the blocky rock mass system stability. In conclusion, automated extraction of ISDRBs can provide accurate quantitative references for rockfall disaster prevention and mitigation design.