GPS-denied
No satellite reference underground means drift cannot be corrected by conventional GNSS.
IROS 2025 · Hangzhou
LiDAR · IMU · UWB · WHEEL
CM-LIUW-
Odometry
Robust and High-Precision LiDAR-Inertial-UWB-Wheel Odometry for Extreme Degradation Coal Mine Tunnels
1 China University of Mining and Technology · 2 Information Institute, Ministry of Emergency Management
Real-world coal mine experiment
Click to inspect ↗
01 · The challenge
When GPS disappears,
geometry can disappear too.
Long, repetitive coal mine tunnels create a rare combination of unavailable global positioning, weak geometric observability, slippery terrain, dust, and limited UWB deployment range.
Feature-poor
Long, straight, repetitive surfaces weaken LiDAR observability and can trigger catastrophic degeneration.
Range-limited
UWB provides global constraints near anchors, but mine-scale coverage is expensive and spatially limited.
02 · The method
One estimator.
Three motion modes.
CM-LIUW-Odometry tightly couples heterogeneous measurements in a 36-dimensional IESKF state, then uses covariance decomposition to decide when wheel constraints should support a degenerating LiDAR–inertial estimate.
Global alignment
LiDAR–inertial estimation receives absolute UWB position and range constraints.
Efficient odometry
LiDAR and IMU carry the estimate while the covariance remains well conditioned.
Wheel-assisted stability
NHC and lever-arm compensated wheel updates suppress drift beyond UWB range.
UWB position + range
Aligns the local SLAM state with the total-station global frame and rejects range outliers.
Wheel + NHC
Adds forward-motion information with nonholonomic constraints and calibrated lever-arm compensation.
Covariance SVD
Finds the direction and severity of weak observability without a fixed scene-specific motion policy.
03 · Real-world results
Stability beyond
the anchor boundary.
Experiments on the CUMT 5 platform compare the complete system against ablations and four state-of-the-art odometry methods in a long, extremely degraded underground tunnel.
Field demonstration
See the estimator move through the tunnel.
Watch the complete IROS 2025 demonstration, including the underground platform and mapping comparison.
Watch on Bilibili
04 · Dataset
Coal mine data,
ready to explore.
The underground ROS bag is publicly available through Baidu Netdisk. Calibration, total-station evaluation files, checksums, mirrors, and standalone dataset terms are still being prepared.
Baidu extraction code idnx
Public bag available
CM-LIUW-CoalMine
Extreme-degradation underground tunnel sequence collected by the CUMT 5 mobile robot.
- LiDAR
- Robosense 32
- IMU
- Xsens-G710
- UWB
- p440 + 4 anchors
- Motion
- Wheel odometry
- Reference
- Total station
- Format
- ROS bag
rosbag · availablecalibration · pendingground truth · pendingchecksum · pending
05 · Citation
Build on our work.
If CM-LIUW-Odometry contributes to your research, please cite the IROS 2025 paper.
@inproceedings{hu2025cmliuw,
author = {Kun Hu and Menggang Li and Zhiwen Jin and
Chaoquan Tang and Eryi Hu and Gongbo Zhou},
title = {{CM-LIUW-Odometry}: Robust and High-Precision
{LiDAR}-Inertial-{UWB}-Wheel Odometry for Extreme
Degradation Coal Mine Tunnels},
booktitle = {2025 IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS)},
year = {2025},
pages = {7139--7146},
doi = {10.1109/IROS60139.2025.11245914}
}