Using Lidar for occupancy grid, depth camera for point cloud

[Spudoom]

Hi,

I currently have a setup with a 3D lidar and a depth camera (D435). From my experiments so far, I have been able to use either the lidar or depth camera to generate the 2D occupancy grid and cloud map at the same time. However, is there a way I can specify to use the lidar for generating the 2D occupancy grid, but the depth camera for the cloud map?

I've tried fusing the lidar point cloud with the D435's point cloud, but the lidar point cloud does not have colour so I can't use them as a single point cloud input for RTABMap.

[matlabbe]

Hi,

With Grid/FromDepth=false, you get occupancy grid created from lidar. Note that in RVIZ (with rtabmap_ros/MapCloud plugin) or rtabmapviz, you can still show the 3D point clouds from the camera.

It is also possible to use rtabmap_ros/map_assembler node (subscribed to mapData output of rtabmap) to reassemble a point cloud with different Grid parameters, like for example using Grid/FromDepth=true to get an occupancy grid from camera at the same time. To be able to do so (when local grids are already created by rtabmap), I just added an option called regenerate_local_grids to ignore local occupancy grids so that they are regenerated with Grid parameters requested. You can see the example in that file.

cheers,
Mathieu