Re: Using Kinect2 and LiDAR to Map and Navigate

Posted by matlabbe on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Using-Kinect2-and-LiDAR-to-Map-and-Navigate-tp6599p6608.html

Hi,

This is a common problem when using only lidar, in particular if it cannot be located at the bottom of the robot to see small obstacles.

The lidar is quite useful for mapping and localization, so I would still feed the lidar to RTAB-Map for better localization. For the map rtabmap is creating, you have two choices:
1) Use lidar (default), and the map will have only obstacles at the height of the robot
2) Use depth image projection (Grid/FromDepth=true), and the map will have the small obstacles but more limited to field of view of the camera

If the small objects are not static, I would go for the first approach, and configure the local costmap from move_base to use both lidar and depth camera cloud to avoid small obstacles. See this config for example. clearing for the lidar could be set to off if you really want to make sure small objects getting off sight of the camera are not cleared. You seem following the similar example. Make sure obstacles_cloud and ground_cloud are connected to move_base node (rqt_graph), there could be a remapping issue. Make sure those topics used in the local costmap are published.

You parameters with lidar are somewhat standard. It will be easier to help if you have more specific problems.

cheers,
Mathieu