Landslide robotics: a prototype for interactive and sustainable geohazard investigation

Landslide studies are currently undergoing a technological revolution facilitated by automated platforms. Collaborative observations and intelligent interpretations based on remote sensing techniques have been integrated into landslide early warning systems, demonstrating advantages in dynamic evolution simulation and probabilistic forecasting. However, extensive labor remains indispensable for ground-based observations to address stochastic natural conditions and obtain accurate quantitative information, particularly in urban zones characterized by frequent and diverse human activities. This study preliminarily investigates the integration of robotics with remote sensing techniques to identify predisposing factors and enhance automation in field investigation of landslides. We propose a prototype of landslide robotics that employs quadruped robots for field investigations, aiming to reduce labor consuming and mitigate potential risks, meanwhile enhance sustainable perceptions and improve interactive automation in geohazard monitoring. Our results indicate that it is practical to digitally characterize the rock slope with meticulous trace cataloging and statistical analysis of semantic components in landform based on high density point cloud data acquired by quadruped robot explorations.