Development of a robot for in situ detection of loess geological information based on machine vision

Abstract

The spatial and temporal distribution of loess geological information and its variations under external disturbances can transparentize and digitize the entire process of the genesis, development, and triggering of geological hazards. However, the lack of reliable detection equipment poses a technical bottleneck to this work. To address this, a specialized detection robot has been developed for exploring the intricate structural defects within 150-mm-diameter geological exploration boreholes, as well as for adapting to optical environments and geological features. Accompanied by an optical environment adaptive control algorithm and equipped with a high-precision industrial camera, the robot captures images of the loess geological information at any position within the borehole. This facilitates intelligent image recognition and provides the necessary conditions for obtaining geological information such as moisture content, porosity and fractures, and interfaces. Indoor and outdoor experimental results demonstrate that these robots have a load capacity exceeding 60 kg, facilitating the integration of other detection instruments. Moreover, within complex loess detection boreholes characterized by localized collapse, collapse, debris, and diameter reduction, the robots not only exhibit stable locomotion with a walking speed of up to 13.18 m/h but also maintain a stable distance of 65 ± 0.1 mm between the industrial camera and the collected images of the borehole wall, within the camera’s depth of field, ensuring stable image brightness and guaranteeing the quality of the captured images. The robots developed in this study provide new technical means and platforms for in situ detection of loess geological information.