Logo U. Osnabrueck Institute of Computer ScienceKnowledge-Based Systems Research Group

6D SLAM with Application in Autonomous Mine Mapping

Merging 3D scans of an old mine

The scanning was done @ CMU. Thanks for the data set.

You may watch the animations:

Meassured and registrated 3D Points

Animation through the whole 3D scene (48 scans):
(animated gif 114 MB) (DivX 53 MB)

merging 3D scans of a abandoned mine

Octree based representation of the 3D point cloud

Animation through the mesh of the 3D scene (points closer than 8 meters have been taken into account for the octree generation):
(animated gif 186 MB) (DivX 53 MB)

Animation of the scan matching process (ICP iterations).

Note: The actual scan matching is done with a global 6D error minimation.

Front view: (animated gif 4 MB) (DivX 6 MB)

Top view: (animated gif 2 MB) (DivX 6 MB)

Viewing hints

For viewing animated gifs you may use IE, Netscape, Mozilla, Konqueror, Pixie. Best viewed with Konqueror or Pixie under Linux. Please download the animated gif before viewing. DivX AVIs can be viewed with MPlayer (Linux) or the Windows Media Player. I suggest viewing the multiple gifs on a fast computer, since there aren't any artefacts.

Acknowledgements

Many thanks to Dirk Hähnel for providing the 3D data.

Also many thanks to Kai Lingemann for converting the data into the file format used in the 3D laser scanner and Kurt3D project.

Paper

Andreas Nüchter, Hartmut Surmann, Kai Lingemann, Joachim Hertzberg, and Sebastian Thrun. 6D SLAM with Application in Autonomous Mine Mapping, in Proceedings IEEE 2004 International Conference Robotics and Automation (ICRA '04), New Orleans, USA, Omnipress, ISBN 0-7803-8233-1, pages 1998 - 2003, April 2004, [Get Paper (PDF)] [HTML version].