Robotiq 2f-140 gripper not moves smoothly


Nowadays I’m trying to control UR5e with Robotiq 2f-140. I think I almost complete the work of integrating the robot and gripper.

However, when I try to close the gripper, elements of the gripper vibrate. I tried after removing all collision parts but, it was still repeated. Also, some revolute joints break away from their position on the link.

(I controlled the left and right base joints separately through GUI.)

How can I fix it???
ur5e_handeye_gripper_v3.usd (11.8 MB)
Here is my USD file. 2F-140 can control with /ur5e/robotiq_arg2f_base_link/right_outer_knuckle_joint and /ur5e/robotiq_arg2f_base_link/left_outer_knuckle_joint.


I handled the vibration problem by changing the option ‘default’ to ‘guide’ atPhysics-Joint-Purpose.

Is it the right approach??

However, I’m still trying to move the gripper faster and, still don’t know why the actuated result angle is related to ‘Damping’ and ‘Stiffnetss’ at Physics-Drive.

It seems the joint limits are actually preventing your system to find a stable pose on high values for the drive target.

I was able to get a stable simulation by changing a bit how the loop was configured.

First thing is set the joint stiffens to higher values - Currently both joint drives have 20.0 for both stiffness and damping, I’d increase it to 200 or 2000.
second right_finger_joint and left_finger_joint limits are blocking the simulation from achieving the final target pose. I removed the joint limits and the simulation became very smooth to me.

See video attached.

1 Like

Thank you!!!
It works well!!!

If someone needs to use UR5e+2f-140+ realsense D435i, use the below USD file.
ur5e_handeye_gripper.usd (11.8 MB)

I have another question. Why do the stiffness(k) and damping(b) constant affect the target (angular) position? I know the mass(m), k, and b are related to the force but I think they cannot affect the result position.


The simulation works as a constrained optimization problem, the stiffness is how strongly it will try to solve for the target position, and the damping is how strongly it will try to solve for the target velocity (usually set to zero, and that’s why it acts as a damping), the lower the value the more relaxed the joint will be and it will settle before the target pose.

1 Like

Oh… Thank you!
I clearly understand.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.