3D location estimation and tunnel mapping of autonomous driving robots through 3D point cloud registration on underground mine rampways

Abstract

• We present a 3D location estimation and tunnel mapping system for autonomous driving robots. • Utilizing 3D LiDAR, we capture and register point clouds for accurate 3D location estimation. • Our system facilitates 3D mapping of underground mine tunnels using point cloud data. In this study, we developed a three-dimensional (3D) location estimation and tunnel mapping system to locate an autonomous robot in the rampway of an underground mine using 3D point cloud registration. A 3D point cloud of the mine tunnel was measured using a 3D light detection and ranging (LiDAR) sensor and registered using the iterative closest point (ICP) algorithm to estimate the 3D pose of the robot. This was combined with two-dimensional LiDAR, inertial measurement unit, and encoder sensors to estimate the 3D trajectory of the robot. Additionally, the 3D tunnel mapping was performed using the 3D trajectory of the robot and the 3D point cloud data of the tunnel. A comparison of the tunnel maps created using conventional surveying equipment and the robot indicated a mapping error of 0.2275 m and localization error of 0.2465 m confirming the excellent overall tunnel mapping and localization performance. The tunnel mapping areas were further compared by selecting areas with relatively high and low ICP matching accuracies; the calculated errors were 0.6186 and 0.2257 m in the areas with low and high accuracies, respectively. Furthermore, the accuracy of the ICP matching tended to be low in areas where the change in the pitch angle of the robot was large.