Example1

In this example, we correct a map (pose graph) constructed by hdl_graph_slam.

Example bag file: hdl_400_dumped.tar.gz (The example file can be reproduced with hdl_graph_slam)

1. Start interactive_slam

rosrun interactive_slam interactive_slam

2. Open graph directory

Menubar -> File -> Open -> New map -> Choose hdl_400_dumped directory

3. Perform automatic loop closing

Menubar -> Graph -> Automatic loop detection

Configure the loop detection parameters as follows and press "Start" button:

Scan matching

• Method: NDT_OMP
• Fitness score thresh: 0.5

Loop detection

• Distance thresh: 20
• Accum distance thresh: 25 or 0.5 (dense loop closing)

Before loop closing:

After loop closing (Accum distance thresh = 25):
(Doubled walls are corrected)

After dense loop closing (Accum distance thresh = 0.5):

4. Save corrected graph

Menubar -> File -> Save -> Save map data -> Choose the destination directory

5. Export point cloud

Menubar -> File -> Save -> Export point cloud -> Choose the destination filename

You can open the exported point cloud with any PCD compliant viewers (e.g., pcl_viewer and CloudCompare)