rtabmap

If there are many TF warnings, you can try to increase robot_state_publisher’s publishing frequency in turtlebot_bringup/launch/includes/robot.launch.xml from 5 to 10 Hz. You can also increase wait_for_transform argument of demo_turtlebot_mapping.launch to 0.2.

默认情况，rtabmap咋mapping模式下，为了在已有地图的模式下进行定位，必须设置incremental参数为false.
By default, rtabmap is in mapping mode. To set in localization mode with a previously created map, you should set the memory not incremental (make sure that arguments don’t contain “–delete_db_on_start” too!):

<launch>
<node name="rtabmap" pkg="rtabmap_ros" type="rtabmap" args="">
   <!-- LOCALIZATION MODE -->
   <param name="Mem/IncrementalMemory" type="string" value="false"/>
</node>
</launch>

map_optimizer

This node is for advanced usage only as it is preferred to use graph optimization already inside rtabmap node (which is the default). See related parameters in rtabmap:

$rosrun rtabmap_ros rtabmap –params | grep Optimize

This node subscribes to rtabmap output topic “mapData” and optimize the graph, then republishes the optimized “mapData”.

This node can be used to optimize the graph outside rtabmap node. The benefice to do that is that we can keep optimized the global map instead of the local map of rtabmap. You can then connect output mapData_optimized to map_assembler to get the optimized grid, proj and cloud maps assembled again. Note that processing time for map optimization using this node is not bounded (which is the case in rtabmap node).

定位模式下，不需要
You could also use your own graph optimization approach instead of this node by modifying poses values of the rtabmap_ros/MapData msg. However, implementing your graph optimization approach inside rtabmap is preferred (inherit Optimizer class and add it here with a new number, then you could select it after using the parameter RGBD/OptimizeStrategy).

When using graph optimization outside rtabmap node, you should set parameters RGBD/OptimizeIterations to 0, RGBD/OptimizeMaxError to 0 and publish_tf to false for rtabmap node.

This node should not be used if rtabmap node is in localization mode. 

5.1 rtabmap_ros/data_odom_sync数据同步

5.2 rtabmap_ros/data_throttle数据节流
Throttle at a specified rate the OpenNI data.

5.7rtabmap_ros/obstacles_detection

This nodelet extracts obstacles and the ground from a point cloud. The camera must see the ground to work reliably. Since the ground is not even, the ground is segmented by normal filtering: all points with normal in the +z direction (+- fixed angle) are labelled as ground, all the others are labelled as obstacles. The images below show an example.

rtabmap参数更改：