Correct mass and center or mass?

ROS
1

By comparing the urdf model of UR5e with the mass properties in this article, I noticed that there were some differences, especially in mass. Which data source is closest to the true values? Is there a more official spec somewhere?

For reference, UR5e link mass comparisons (urdf on the right):
Link1: 3.71 vs 3.7
Link2: 8.058 vs 8.393
Link3: 2.846 vs 2.275
Link4: 1.37 vs 1.219
Link5: 1.3 vs 1.219
Link6: 0.365 vs 0.1879

We are using the urdf model in a simulation, so it is important that the mass properties are correct.

2

We are currently fixing this on the melodic-devel-staging branch. See also this PR related to this. However, I think I indeed missed to update the masses there, as well.

3

Thanks mauch!

Currently trying to use the updated values in our project.

Some things I noticed in physical_parameters.yaml (UR5e) of the PR:

  • upper_arm_cog.x is set to -0.17 but comment says that model.x - upper_arm_length, which with values inserted leads to x = 0.2125 - (--0.425) = -0.2125, i.e. not -0.17. Is this a mistake or is there a reason for it being set to -0.17? From what values should ‘upper_arm_length’ and ‘forearm_length’ be read?

  • Almost all dh_parameters are slightly different from UR website and default_kinematics.yaml. Very small difference though. Should these in theory be set the same?

Somewhat out of scope of this forum, but:

  • In our project we need to use .urdf files. I previously converted the UR5e root xacro file (of the kinetic branch) to urdf using the following command:

    rosrun xacro xacro -o "Path\ur5e_robot.urdf" "Path\ur5e_robot.urdf.xacro"

    …but in the melodic branch the file structure has changed quite a bit and the root file for UR5e seems to now be load_ur5e.launch. Do you have any hint on how to generate an UR5e urdf file from this? Or do I have to create my own UR5e xacro root file manually first? (I’m actually currently doing this on Windows)

4

Noticed that the center of mass from UR for link6 on UR5e is at the very end of the link ( [0, 0, -0.001159] ). This seems strange, and even physically impossible, I would assume it to be more “inside” the actual link.

5

I’ll come back to that… However, best we discuss this further on GH.