Scholarlink

<p>In underground mines, enclosed tunnels, and other Global Navigation Satellite System (GNSS)-denied environments during coal exploration, geological obstructions, dust interference, dynamic obstacles, and extreme lighting variations pose severe challenges to accurate localization. This paper proposes LiVIC-EKF, a novel multi-sensor fusion framework that integrates light detection and ranging (LiDAR), binocular cameras, and inertial measurement unit (IMU) to achieve robust six-degree-of-freedom pose estimation. The approach enhances the LiDAR odometry and mapping (LOAM) algorithm to efficiently extract LiDAR geometric features, enabling high-precision point cloud registration and robust environmental mapping for low-texture coal mine environments. By tightly coupling visual-inertial odometry (VIO) using the VINS-Fusion framework, the proposed method mitigates motion blur and scale ambiguity through IMU pre-integration and sliding-window nonlinear optimization, thereby improving robustness to rapid motion and lighting variations. A key advancement is the dynamic covariance-adaptive mechanism, which adaptively weights the contributions of LiDAR and VIO based on real-time environmental conditions, such as dust density or the success rate of visual feature tracking, using a sigmoid function to ensure smooth state estimation transitions and suppress outliers. Offline sensor calibration and a hierarchical fault detection strategy further enhance reliability by addressing systematic errors and sensor asynchrony. Experimental validation on the KITTI dataset demonstrates that LiVIC-EKF achieves a mean absolute pose error (APE) of 2.60&#xa0;ms and a standard deviation (SD) of 1.66&#xa0;ms, outperforming state-of-the-art methods like A-LOAM, FAST-LIO2, and VINS-Fusion. This study provides a practical and robust localization solution for coal exploration, where GNSS unavailability and harsh conditions demand reliable autonomous navigation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-sensor information fusion methods for coal mine exploration in GNSS-denied scenarios

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月10日

查看全文
  • Shuai Wang,
  • Yuntao Liang,
  • Shengtao Liao,
  • Shuai Xiao,
  • Liao Wang

摘要

In underground mines, enclosed tunnels, and other Global Navigation Satellite System (GNSS)-denied environments during coal exploration, geological obstructions, dust interference, dynamic obstacles, and extreme lighting variations pose severe challenges to accurate localization. This paper proposes LiVIC-EKF, a novel multi-sensor fusion framework that integrates light detection and ranging (LiDAR), binocular cameras, and inertial measurement unit (IMU) to achieve robust six-degree-of-freedom pose estimation. The approach enhances the LiDAR odometry and mapping (LOAM) algorithm to efficiently extract LiDAR geometric features, enabling high-precision point cloud registration and robust environmental mapping for low-texture coal mine environments. By tightly coupling visual-inertial odometry (VIO) using the VINS-Fusion framework, the proposed method mitigates motion blur and scale ambiguity through IMU pre-integration and sliding-window nonlinear optimization, thereby improving robustness to rapid motion and lighting variations. A key advancement is the dynamic covariance-adaptive mechanism, which adaptively weights the contributions of LiDAR and VIO based on real-time environmental conditions, such as dust density or the success rate of visual feature tracking, using a sigmoid function to ensure smooth state estimation transitions and suppress outliers. Offline sensor calibration and a hierarchical fault detection strategy further enhance reliability by addressing systematic errors and sensor asynchrony. Experimental validation on the KITTI dataset demonstrates that LiVIC-EKF achieves a mean absolute pose error (APE) of 2.60 ms and a standard deviation (SD) of 1.66 ms, outperforming state-of-the-art methods like A-LOAM, FAST-LIO2, and VINS-Fusion. This study provides a practical and robust localization solution for coal exploration, where GNSS unavailability and harsh conditions demand reliable autonomous navigation.