mapData visualization over remote Rviz

[rakuwaku]

Hi,
 
  For my purposes I want to rely on mapData rather than the cloud_map for visualization in rviz. I want to teleoperate the robot and use the SLAM map to give the user a good perspective of the environment. I'm just not getting that with point clouds.
However, mapData is much more bandwidth demanding. If I increase Grid/CellSize I can reduce the bandwidth, but I'd rather not sacrifice that.

Are there any settings that I can tweak? I'm not seeing huge gains when I tweak decimation for mapData in the Rviz settings.

 

[matlabbe]

Hi,

Grid/CellSize will have an impact on cloud_map for the bandwidth, not mapData. mapData contains always all data from the latest node added to map. By default, depth and rgb images are recorded compressed at input resolution. If you want to reduce bandwidth with mapData, 2 choices:

1) Mem/BinDataKept=false, rgb and depth images won't be saved and published. Probably not the option you want.

2) Mem/ImagePostDecimation, rgb and depth images will be saved decimated. mapData will then use less bandwidth because the images will be already decimated. You may want to set decimation in rviz's MapCloud plugin to 1 to see all points.

Note that cloud_map bandwidth will increase over time as the map gets bigger. It is not the case for mapData, which should be relatively constant (only the graph is growing linearly, which can be neglected)

cheers,
Mathieu