Revision history [back]

The navigation costmaps used in the nav stack work on both, point clouds and laser scans. I guess using big point clouds produced by e.g. the kinect can slow down the costmaps because lots of points have to be added at a pretty high rate. That might be the reason for using laser scans generated from point clouds instead of using the complete point cloud although it definitely contains more information.

Nodes for localization and map building, e.g. amcl and gmapping, need laser scans however.