Accurate Real-Time Localization Estimation in Underground Mine Environments Based on a Distance-Weight Map (DWM)

Area of Science:

• Robotics
• Geospatial Intelligence
• Mining Engineering

Background:

• Precise localization is crucial for intelligent underground mining but faces challenges like GPS unavailability, poor lighting, visual aliasing, dust, and water.
• Existing localization methods struggle in complex underground mine environments.

Purpose of the Study:

• To develop a high-precision, real-time, infrastructure-free underground localization method using 3D LIDAR.
• To introduce a novel error correction technique for point cloud mapping in underground environments.

Main Methods:

• Constructed an underground mine environment map using Generalized Iterative Closest Point Simultaneous Localization and Mapping (GICP-SLAM).
• Proposed Inverse Distance Weighting (IDW) for error correction, creating a Distance-Weight Map (DWM).
• Combined point cloud frame matching and DWM matching within an unscented Kalman filter for fusion-based localization.

Main Results:

• Achieved a spatial localization error of 4 cm.
• Demonstrated a processing time of 60 ms per frame for real-time performance.
• Validated the method in four distinct underground mine scenes.

Conclusions:

• The developed algorithm enables low-drift, real-time localization in challenging underground mine environments.
• The Distance-Weight Map (DWM) approach effectively corrects errors in point cloud mapping.
• This method significantly advances the potential for unmanned and intelligent underground mining operations.