Good day everyone,

I have just signed up here on the forum after having a very strange behaviour with gazebo simulation. Since this problem can also be from my xacro file, I have also posted this question on ROS forum: https://answers.ros.org/question/376336/xacro-strangest-behaviour-with-inertia/ (awaiting moderation).

The issue is the following:

I was designing a robot with xacro and started seing the strangest behaviours from the component. I have a component, which is rotated 90 degrees in the roll axis. The inertia was also rotated (as the visual and collision), but, analysing in gazebo, I see that the inertia ends up not being rotated, as shown in the following image:

I thought this behaviour to be very weird and so I tried with a different inertia, this time not the one taken from the CAD but considering that the inertia from the object was represented by a cuboid. So, I used the following inertia (which is very close to the one given by the CAD tool for the object):

<inertia ixx="8.5474390625e-4" ixy="0" ixz="0" iyy="8.5474390625e-4" iyz="0" izz="1.7094878125e-3" />

This time, gazebo couldn't even display the inertia, as shown here:

Finally, with a try and error approach, i changed the izz moment of inertia to 1.5e-3, resulting in:

<inertia ixx="8.5474390625e-4" ixy="0" ixz="0" iyy="8.5474390625e-4" iyz="0" izz="1.5e-3" />

This time, i got the inertia to be displayed correctly: .

Everything seems correct to me in all the cases (I have only changed the component inertia between trials) so, what may be causing this issue?