Posted by
andreacelani on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Mapping-and-Navigation-with-Intel-R200-depth-camera-tp2583.html
Hi Mathieu,
I'm trying to activate the obstacle_detection nodelet and fed it to the local costmap but I'm not able to map correctly the input/output of the nodelet.
I have an intel R200 depth camera and visual odometry activated.
Everything is fine with visual odometry, so the sensor and other stuff are working great.
I disabled the depthimage_to_laserscan nodelet and now the /map takes rtabmap/proj_map data
I have modified my local costmap params file to include obstacle_detection points:
***
observation_sources: point_cloud_sensor
# assuming receiving a cloud from rtabmap_ros/obstacles_detection node
point_cloud_sensor: {
sensor_frame: base_footprint,
data_type: PointCloud2,
topic: /planner_cloud,
expected_update_rate: 0.5,
marking: true,
clearing: true,
min_obstacle_height: -99999.0,
max_obstacle_height: 99999.0}
***
Then I activated the nodelet in this way on my main .launch file:
***
<node pkg="nodelet" type="nodelet" name="obstacles_detection" args="load rtabmap_ros/obstacles_detection nodelet">
<remap from="cloud" to="rtabmap/cloud_map"/>
<remap from="obstacles" to="/planner_cloud"/>
</node>
***
But when I start the launch file the obstacle_detection nodelet is disconneted from everything else.
Basically it's a newbie question related to mapping in ROS.
I understand the output mapping, for example:
<remap from="obstacles" to="/planner_cloud"/> will rename the obstacles stream as /planner_cloud stream.
But it's not clear to me the input remapping like:
<remap from="cloud" to="rtabmap/cloud_map"/>
I suppose input stream of obstacle_detection called cloud (subscribed topic) is remapped as rtabmap/cloud_map (published topic of rtabmap) or it's the opposite?
I'm missing something.
Moreover I would like to understand the difference between proj_map topic from rtabmap and the obstacles topic from obstacle_detection nodelet.
It's not clear to me they seem very similar in terms of processed data.
Thank you
Andrea