CM-LIUW-Odometry: Robust and High-Precision LiDAR-Inertial-UWB-Wheel Odometry for Extreme Degradation Coal Mine Tunnels

TL;DR

The paper addresses GPS-denied, feature-poor underground coal mine SLAM by proposing CM-LIUW-Odometry, a multimodal SLAM framework that tightly fuses LiDAR-Inertial Odometry with UWB absolute positioning and integrates wheel odometry under nonholonomic constraints. It leverages an Iterated Error-State Kalman Filter to fuse multiple sensor modalities and introduces a covariance PCA-based degradation detection with adaptive motion mode switching to maintain accuracy beyond the UWB range. Key contributions include global-position alignment via UWB, extended robustness through wheel odometry with lever-arm compensation, and an adaptive mechanism that switches among LIU, LIO, and LIW modes. Real-world experiments in degraded tunnels demonstrate superior accuracy and robustness compared with state-of-the-art methods, and the authors provide open-source code for the robotics community.

Abstract

Simultaneous Localization and Mapping (SLAM) in large-scale, complex, and GPS-denied underground coal mine environments presents significant challenges. Sensors must contend with abnormal operating conditions: GPS unavailability impedes scene reconstruction and absolute geographic referencing, uneven or slippery terrain degrades wheel odometer accuracy, and long, feature-poor tunnels reduce LiDAR effectiveness. To address these issues, we propose CoalMine-LiDAR-IMU-UWB-Wheel-Odometry (CM-LIUW-Odometry), a multimodal SLAM framework based on the Iterated Error-State Kalman Filter (IESKF). First, LiDAR-inertial odometry is tightly fused with UWB absolute positioning constraints to align the SLAM system with a global coordinate. Next, wheel odometer is integrated through tight coupling, enhanced by nonholonomic constraints (NHC) and vehicle lever arm compensation, to address performance degradation in areas beyond UWB measurement range. Finally, an adaptive motion mode switching mechanism dynamically adjusts the robot's motion mode based on UWB measurement range and environmental degradation levels. Experimental results validate that our method achieves superior accuracy and robustness in real-world underground coal mine scenarios, outperforming state-of-the-art approaches. We open source our code of this work on Github to benefit the robotics community.

Figures (8)

• Figure 1: System overview.
• Figure 2: Lever arm compensation between the IMU and the wheel odometer coordinate system. When the vehicle turns, the IMU and the wheel odometer have different velocities, $^I\mathbf{v}$ and $^W\mathbf{v}$, respectively. Therefore, the wheel odometer velocity $^W\mathbf{v}$ needs to be transformed from the wheel odometer coordinate system $\mathbf{O}_W$ to the IMU coordinate system $\mathbf{O}_I$ using their extrinsic parameters $^I\mathbf{R}_W$ and $^I\mathbf{p}_W$ for fusion.
• Figure 3: Schematic diagram of degradation detection and adaptive motion mode switching. (a) When the robot is within the measurement range of the UWB positioning system, the system uses the LIU mode for localization, and the covariance ellipsoid is uniformly distributed; (b) After the robot exits the UWB measurement range, the system switches to LIO mode. Due to the loss of the X-axis constraint, the covariance ellipsoid expands significantly along the X-axis direction, leading to system degradation; (c) After detecting degradation in the X-axis direction, the system introduces the wheel odometer constraint $\mathbf{z}_{W}^{k+1}$ and switches to LIW mode, reducing the covariance ellipsoid in the degraded direction; (d) Shows the UWB and robot positions in the global coordinate system $G$ and the UWB measurement range $\mathcal{R}_U$.
• Figure 4: Field experiment environment in the underground coal mine tunnel. (a) Shows the deployment locations of UWB, CUMT_5 robot, and total station in the underground coal mine tunnel. (b) Shows the sensor layout of LiDAR, UWB, IMU, and wheel odometer on CUMT_5. (c) and (d) Show the environment and texture details of the outer tunnel and inner tunnel, respectively. (e) Shows the locations of UWB, CUMT_5 robot, and total station in the underground coal mine tunnel, as well as the maximum measurement range $\mathcal{R}_U$ of the UWB positioning system.
• Figure 5: Mapping performance of (a) Ours, (b) Ours w/o wheel, (c) Ours w/o uwb, (d) Ours w/o wheel & uwb, (e) FAST-LIO2, (f) DLIO, (g) IG-LIO, and (h) LIO-SAM in an underground coal mine environment. Significant degradation in the inner tunnel is observed for Ours w/o wheel, DLIO, IG-LIO, and LIO-SAM. Ours w/o uwb & wheel and FAST-LIO2 fail entirely, while Ours and Ours w/o wheel successfully navigate to the tunnel endpoint.