BA-CLM: A Globally Consistent 3D LiDAR Mapping Based on Bundle Adjustment Cost Factors.

Area of Science:

• Robotics
• Computer Vision
• Geomatics

Background:

• Globally consistent, high-precision mapping is critical for mobile robot applications.
• Current optimization-based methods often lead to map inconsistencies by not directly optimizing scene structure.

Purpose of the Study:

• To present a novel 3D LiDAR mapping framework, BA-CLM, addressing inconsistencies in existing methods.
• To introduce a multivariate LiDAR bundle adjustment (LBA) cost factor for improved robot pose constraint.

Main Methods:

• Developed a 3D LiDAR mapping framework (BA-CLM) utilizing LiDAR bundle adjustment (LBA) cost factors.
• Proposed a multivariate LBA cost factor derived from a multi-resolution voxel map.
• Applied LBA cost factors for both local and global map optimization within the framework.

Main Results:

• The BA-CLM framework demonstrated accurate trajectory estimation.
• Achieved consistent mapping results across multiple public and self-collected 3D LiDAR datasets.
• Validated the effectiveness of the proposed multivariate LBA cost factor.

Conclusions:

• The proposed BA-CLM framework effectively overcomes the limitations of existing methods, enabling consistent and precise 3D LiDAR mapping.
• The novel LBA cost factor contributes significantly to the accuracy of robot pose estimation and overall map quality.
• This framework offers a robust solution for mobile robot navigation and scene understanding.