A Pointcloud Registration Framework for Relocalization in Subterranean Environments

TL;DR

The paper tackles relocalization for autonomous robots operating in GPS-denied subterranean environments, where noise, dust, and irregular surfaces hinder localization. It introduces a robust three-stage point cloud registration pipeline that starts with ISS keypoint detection and FPFH descriptor matching to obtain an initial transform via RANSAC, followed by NDT-based refinement and a final ICP pass. Key contributions include a computationally efficient, multi-stage framework tailored to underground conditions and validated on both simulated mine environments and real-world mine data, showing improved robustness and accuracy. The approach has practical implications for reliable navigation in mines and tunnels, enabling safer exploration, inspection, and rescue operations in challenging settings, with $N(oldsymbol{x}; oldsymbol{Cmu}, oldsymbol{CSigma})$ modeling in-voxel distributions and likelihood-based optimization guiding alignment.

Abstract

Relocalization, the process of re-establishing a robot's position within an environment, is crucial for ensuring accurate navigation and task execution when external positioning information, such as GPS, is unavailable or has been lost. Subterranean environments present significant challenges for relocalization due to limited external positioning information, poor lighting that affects camera localization, irregular and often non-distinct surfaces, and dust, which can introduce noise and occlusion in sensor data. In this work, we propose a robust, computationally friendly framework for relocalization through point cloud registration utilizing a prior point cloud map. The framework employs Intrinsic Shape Signatures (ISS) to select feature points in both the target and prior point clouds. The Fast Point Feature Histogram (FPFH) algorithm is utilized to create descriptors for these feature points, and matching these descriptors yields correspondences between the point clouds. A 3D transformation is estimated using the matched points, which initializes a Normal Distribution Transform (NDT) registration. The transformation result from NDT is further refined using the Iterative Closest Point (ICP) registration algorithm. This framework enhances registration accuracy even in challenging conditions, such as dust interference and significant initial transformations between the target and source, making it suitable for autonomous robots operating in underground mines and tunnels. This framework was validated with experiments in simulated and real-world mine datasets, demonstrating its potential for improving relocalization.

Figures (5)

• Figure 1: Flowchart of the relocalization framework for subterranean environments, illustrating the steps of denoising, downsampling, keypoint selection, descriptor computation, and transformation refinement.
• Figure 2: Model mine environment used in the ROS/Gazebo takaya2016simulation simulator.
• Figure 3: Pointclouds before and after registration using the full framework. The source pointcloud is in magenta and the target pointcloud is in green.
• Figure 4: Both test environments: the physically simulated coalmine (left) and the actual limestone mine (right).
• Figure 5: Point cloud registration results: (Top) Simulation mine before and after registration. (Bottom) Limestone mine before and after registration. The source pointcloud is in magenta and the target pointcloud is in green.