Deformation Detection Method Of Coal Mine Rescue Shaft Based On RGB-D And Lidar Sensors

In recent years,mine accidents occur frequently,causing heavy casualties.In the case of light collapse and shallow burial depth,manual obstacle clearance can be used for rescue;in most cases,it is difficult for the trapped people to bury deeply,and the rescue shaft provides a new life channel for the trapped people.However,seepage and geological movement and other factors will lead to the deformation of the rescue well,affect the safety of the rescue work,and make the application of the rescue cabin face great risks.Therefore,in the rescue process,it is urgent to carry out real-time deformation detection of the mine to ensure the safety of the difficult rescue work.According to the characteristics of small diameter,smooth wall and small deformation of the rescue shaft,rgb-d and lidar sensors are used to sense the deformation of the shaft in real time under the conditions of light and dark.RGB-D sensor and slam technology are used to build a real-time 3D model of the rescue shaft wall,and the deformation detection is carried out by vision.Lidar sensor is used to make up for the deformation detection under the condition of no light emergency.The indoor rescue shaft simulation platform is built by using NVIDIA Jetson TX2,Intel realsense D435 i and RPLIDAR A1 constructs the environment perception platform,uses v SLAM and laser slam method to detect the wellbore deformation state,uses ball pivoting algorithm to reconstruct the wellbore point cloud,and uses MATLAB and meshlab to process and recognize the wellbore point cloud deformation.On this basis,the indoor fusion experiment is designed to explore the functional relationship between deformation measurement results and deformation measurement errors of vision sensor and lidar in the case of small change of fixed distance,and then the expansion and scaling coefficients under different measurement results are determined,and a multi-sensor data fusion algorithm based on the expansion and scaling coefficient is proposed.The results show that the RGB-D SLAM based on depth camera can effectively construct the 3D map of shaft and realize the shaft wall deformation detection on the visual level under the condition of light,and lidar sensor can realize the shaft wall deformation detection through infrared laser ranging under the condition of no light.The average error of 1cm,2cm and 4cm deformation recognition is 57%,28.5%,26.74% and 70%,24.5%,8.25% respectively.Through the indoor fusion experiment,the piecewise linear interpolation function between the deformation measurement results and the deformation measurement error of the two sensors is explored.The slopes of the four functions are-10.0952,-8.1039,-10.200,-21.300 respectively.Based on the experimental results,the lidar measurement results and visual inspection results in shaft lining are scaled and expanded.After scaling and expansion,the deformation detection errors of 1cm,2cm and 4cm are 6%,-17.5%,-3.5% and 45%,12.5% and 4.725% respectively.Compared with the original measurement results,the deformation measurement accuracy is effectively improved.By fusing the two sensors with different weights,the error of deformation detection is reduced to 13.8%,2.5% and 1.125%.Using RGB-D and Lidar sensors provides a new method for the environment perception of the rescue shaft and the safety work of the rescue cabin,and provides a new idea for the application of slam technology.