| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Hi! I ran the following demo for rtab-map ROS without changing anything:
$ roslaunch rtabmap_ros demo_robot_mapping.launch rviz:=true rtabmapviz:=false
$ rosbag play --clock demo_mapping.bag
In rviz, when I change the PointCloud2 from /voxel_cloud to /rtab_map/cloud_map, I get a point cloud of a 2D map with only red dots. Is /rtab_map/cloud_map not supposed to be a 3D point cloud?
When subscribing to a laser scan, by default it will generate a 2D map. Set Grid/FromDepth to true to use RGB-D camera to generate the map instead.
I see - however, if I set Grid/FromDepth to true, the occupancy grid then changes to be less accurate than if Grid/FromDepth is false. Is it possible to set Grid/FromDepth=true so that the cloud map is a 3D map, yet also generate an occupancy grid like this?
This is because the camera is seeing the ceiling, thus creating obstacles. If you want to get both, you have to reassemble the cloud with map_assembler:
rosrun rtabmap_ros map_assembler mapData:=/rtabmap/mapData _regenerate_local_grids:=true --Grid/FromDepth true
I see, is it because generating the 2D occupancy grid projects the point cloud to the x-y plane, so parts of the ceiling become obstacles? Thank you so much for your help!
Asked: 2021-03-11 00:41:59 -0500
Seen: 504 times
Last updated: Mar 11 '21
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
