Robust and Fast Point Cloud Registration for Robot Localization Based on DBSCAN Clustering and Adaptive Segmentation
2024

Fast and Accurate Point Cloud Registration for Robot Localization

publication Evidence: high

Author Information

Author(s): Liu Haibin, Tang Yanglei, Wang Huanjie, Rizos Chris

Primary Institution: College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing, China

Hypothesis

The proposed CSNDT algorithm improves the robustness and accuracy of point cloud registration compared to traditional methods.

Conclusion

The CSNDT algorithm significantly outperforms traditional methods in robustness, matching efficiency, and accuracy across various environments.

Supporting Evidence

  • The CSNDT algorithm maintains a high success rate in point cloud matching even with significant offsets.
  • CSNDT shows a 54.1% improvement in computational efficiency compared to the NDT algorithm.
  • Experimental results indicate that CSNDT achieves superior accuracy in environments with linear features.

Takeaway

This study shows a new way to help robots understand where they are by making sense of 3D point clouds better and faster.

Methodology

The study uses an improved DBSCAN algorithm for clustering and adaptive segmentation of point clouds to enhance registration accuracy.

Limitations

The method may struggle with complex environments featuring significant curvature in obstacles.

Digital Object Identifier (DOI)

10.3390/s24247889

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication