Bridge point cloud packed

[nkoenig]

See: https://bitbucket.org/ignitionrobotics/ign-msgs/pull-requests/144 and https://bitbucket.org/ignitionrobotics/ign-sensors/pull-requests/78/output-a-pointcloudpacked-message/diff

[chapulina]

I tried adding a demo in on branch packed_demo using the new message but the point cloud is not showing on RViz. The ROS message data is suspiciously short:

header:
  seq: 105
  stamp:
    secs: 10
    nsecs: 500000000
  frame_id: "model_with_lidar/link/gpu_lidar"
height: 1
width: 1
fields:
  -
    name: "x"
    offset: 0
    datatype: 7
    count: 1
  -
    name: "y"
    offset: 4
    datatype: 7
    count: 1
  -
    name: "z"
    offset: 8
    datatype: 7
    count: 1
is_bigendian: False
point_step: 12
row_step: 12
data: [0, 0, 128, 127, 0, 0, 128, 255, 0, 0, 192, 255]
is_dense: True

So is the Ignition data:

header {
  stamp {
    sec: 13
    nsec: 500000000
  }
  data {
    key: "frame_id"
    value: "model_with_lidar::link::gpu_lidar"
  }
}
field {
  name: "x"
  datatype: FLOAT32
  count: 1
}
field {
  name: "y"
  offset: 4
  datatype: FLOAT32
  count: 1
}
field {
  name: "z"
  offset: 8
  datatype: FLOAT32
  count: 1
}
height: 1
width: 1
point_step: 12
row_step: 12
data: "\000\000\200\177\000\000\200\377\000\000\300\377"
is_dense: true

I'm still looking into what could be going on. It could be a problem with my local setup.

[nkoenig]

Your width and height are 1. Is that intentional?

[chapulina]

Oh I hadn't noticed that detail on the message. I'm running this example world. The laser scans are going through fine. It could be an issue with my setup, I'll try your updated branches soon.

[chapulina]

The GPU lidar works for me with the new branches. The depth cloud looks mirrored though. This is using the demo on #29 :

[nkoenig]

The mirror should be fixed: https://bitbucket.org/ignitionrobotics/ign-sensors/commits/3f70f5615bd01cc8a7639d80500a065e20858d5e?at=ign-sensors2

[nkoenig]

Are you having trouble with the depth_image in rviz? I see all black.

[chapulina]

Are you having trouble with the depth_image in rviz? I see all black.

You need this fix, it hasn't been released yet: ros-visualization/rviz#1378

[nkoenig]

Ahh, thanks.

[chapulina]

Works for me. We need to wait for a couple of releases before merging. And some basic tests would also be nice.

[chapulina]

Since the messages are so similar, it shouldn't be a lot of trouble to add the reverse conversion, and some tests with some dummy data.

[nkoenig]

Add other direction and tests.