Inertial parameters mismatch between URDF and official manual of UR5e

Hello,

I am using UR5e robot for a project where I need to develop a system model and simulate it. While I was studying the inertial parameters I found mismatch in the values of mass, centre of mass, and Inertia of joints between the URDF files and official guide. Can you please tell me which one is correct?

Official manual from website (https://www.universal-robots.com/articles/ur/application-installation/dh-parameters-for-calculations-of-kinematics-and-dynamics/)

From URDF file

[Jishnu Subramonian]

Hi Srinivas,

The MATLAB based urdf file for UR5e reflects the values from ROS-Industrial GitHub,

universal_robot/ur_description/config/ur5e/physical_parameters.yaml at noetic-devel · ros-industrial/universal_robot · GitHub.

We would like to understand your use-case better,

Thanks

[Srinivasan]

Hey, Thanks for answering! I am trying to create a MPC controller for the UR5e manipulator. I wanted to ensure when i model the robot system it is closer to the actual robot. I already saw the source you have mentioned. I want to know the reason for this difference. Is the frame of reference assumed different for the URDF and the values mentioned in official manual? There is no information regarding the same.

Answer 1

[Hannes Daepp]

Thanks for reporting this issue. As others have stated, the discrepancy occurs because the MATLAB version is sourced from the ROS-Industrial repository, and the values were incorrect there. The issue has since been reported and has recently (last week) been updated in the ROS2 repository, with ROS1 forthcoming:

Mass of UR5e are incorrect · Issue #186 · UniversalRobots/Universal_Robots_ROS2_Description · GitHub

You an always find the source used by the robot library by checking the Source property of the info struct, i.e. using the following code:

[robot,info]=loadrobot("universalUR5e");
info.Source

ans = "https://github.com/fmauch/universal_robot/tree/1.2.1/ur_description"

Given the change to the source repository, we will work to update the corresponding values in the robot library (loadrobot) in a near-term update.

Answer 2

[Kautuk Raj]

I can understand that you are trying to resolve discrepancies between the inertial parameters of the “UR5e robot” as found in the URDF files and those listed in the official manual.

The URDF file for the “UR5e robot” in MATLAB is based on the values provided by the ROS-Industrial GitHub repository. Specifically, these parameters are sourced from the ‘physical_parameters.yaml’ file located at the following repository: https://github.com/ros-industrial/universal_robot/blob/noetic-devel/ur_description/config/ur5e/physical_parameters.yaml

More information on handling a “UR5e Cobot” from Universal Robots can be found here in the documentation: https://www.mathworks.com/help/releases/R2023a/supportpkg/urseries/ug/getting-started-controlling-ur5e.html

I hope this clarifies the source of the parameter values in the URDF file.

[Srinivasan]

Hey! Thanks for answering.

I already saw the source you have mentioned. I want to know the reason for this difference. Is the frame of reference assumed different for the URDF and the values mentioned in official manual? There is no information regarding the same.

[Hannes Daepp]

Regarding the comment on inertia matrices, of course a zero inertia matrix doesn't make any sense. But if you read the page, you can see that the post includes the following disclaimer:

It should be noted that the inertia matrix values for UR3e and UR5e are negligible due to their smaller size.

Questions about the inertias and frame of reference in the official manual are best directed to UR.