You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have been working with Reachy's URDF file in conjunction with the Pinocchio library for torque estimation within the joints. However, when using Pinocchio's Recursive Newton-Euler Algorithm (RNEA) alongside Reachy's URDF model, I get incredibly high values for the torque - on the scale of 106. I was getting these values both when running the algorithm directly on the Reachy, and when I used the URDF file on my own machine. A small sample code for testing and the obtained output have been given below:
These extremely high values seemed quite unreasonable for the values of position, velocity, and acceleration provided. However, when digging further through the URDF files, I came across the 'reachy.URDF.xacro' file, which, when compared to the original URDF file, seems to have significantly different inertia values for the following links:
r_shoulder_x
l_shoulder_x
l_gripper_thumb
The original URDF file contains extremely high inertia values for these links, whereas the 'reachy.URDF.xacro' file seems to have much smaller values. The high inertia values are present in all of the other URDF files I have seen as well (including reachy_2023). However, when I replace these high inertia values with the ones I found in 'reachy.URDF.xacro', I get the following results with the modded URDF file:
The torques I get from this modified URDF seem to be much more reasonable. I wanted to basically ask if these very high inertia values in the URDF files are correct, or if the values present in 'reachy.URDF.xacro' are the correct ones, because the latter ones seem to give much more reasonable results.
I'd like to also note here that I believe I am using Reachy 2021, and this discrepancy in inertia values was present on the files on our actual Reachy robot as well. Since the reachy_description GitHub repository has been archived, and the high inertia values seem to be present in reachy_2023 as well, I thought I'd open the issue here.
I would be highly grateful for any help in this matter. Thank you.
The text was updated successfully, but these errors were encountered:
Thanks for letting us know!
The physical model of the arm almost did not change between the 2021 and 2023, so you can use the value from the reachy_description package.
I'll let the issue open so we update the URDF with the correct values!
Hi,
I have been working with Reachy's URDF file in conjunction with the Pinocchio library for torque estimation within the joints. However, when using Pinocchio's Recursive Newton-Euler Algorithm (RNEA) alongside Reachy's URDF model, I get incredibly high values for the torque - on the scale of 106. I was getting these values both when running the algorithm directly on the Reachy, and when I used the URDF file on my own machine. A small sample code for testing and the obtained output have been given below:
These extremely high values seemed quite unreasonable for the values of position, velocity, and acceleration provided. However, when digging further through the URDF files, I came across the 'reachy.URDF.xacro' file, which, when compared to the original URDF file, seems to have significantly different inertia values for the following links:
The original URDF file contains extremely high inertia values for these links, whereas the 'reachy.URDF.xacro' file seems to have much smaller values. The high inertia values are present in all of the other URDF files I have seen as well (including reachy_2023). However, when I replace these high inertia values with the ones I found in 'reachy.URDF.xacro', I get the following results with the modded URDF file:
The torques I get from this modified URDF seem to be much more reasonable. I wanted to basically ask if these very high inertia values in the URDF files are correct, or if the values present in 'reachy.URDF.xacro' are the correct ones, because the latter ones seem to give much more reasonable results.
I'd like to also note here that I believe I am using Reachy 2021, and this discrepancy in inertia values was present on the files on our actual Reachy robot as well. Since the reachy_description GitHub repository has been archived, and the high inertia values seem to be present in reachy_2023 as well, I thought I'd open the issue here.
I would be highly grateful for any help in this matter. Thank you.
The text was updated successfully, but these errors were encountered: