ROS laserscan messages wrong

[chr1s]

Hello,

I have a problem with the ROS laserscan messages published by Isaac Sim.
As you can see in the following screenshot, the laserscan data published as a pointcloud are correct,
while the messages published as laserscan messages are wrong / moving around.

This only happens when the rotation rate of the lidar is not zero. The exact behavior depends on the rotation rate of the lidar. For low rotation rates the points move around, for higher rotation rates they stay at the same location, which however is turned in comparison to the pointcloud.
I know of the old bug where the rotation rate had to be set to zero, but as far as i know this was fixed in version 2021.2.
Attached you find a simple .usd file which reproduces the error.
LidarSample.usd (17.2 KB)

In a more complex environment i also had the issue that points were only present in angles from 0 to 180 degrees and the points from 180 to 360 degrees were mirrored into the other half of the circle. However I wasn’t able to reproduce this in the sample file.

Thanks for your help!

Kind regards
Chris

[Hammad_M]

Thanks @chr1s I’ve filed a bug, will take a look

[Hammad_M]

@chr1s , if you only enable point cloud publishing, or laserscan publishing, they independently work correctly right? The issue only manifests when both are enabled?

[chr1s]

@Hammad_M Yes, that is correct. If i start the simulation with only one of them enabled it works.
I also noticed that if I stop the pointcloud publishing during simulation, the rotation of the laserscan messages stops, but just wherever it is at that moment.

Thanks for taking a look into it!

[Hammad_M]

I believe we’ve fixed issues with the LaserScan and PCL publishers running concurrently. They will be part of the next release.