I’ve tried follow_target_with_rmp.py and follow_target_with_ik.py and find that the converged end effector position has offset with target position. The offset is even more than 1cm, which is not acceptable for position control.

@newuhe Sorry for the slow response. Can you confirm that you’re seeing this for the Franka examples (as opposed to UR10 or DOFBot)?

I’m sure its follow_target_with_rmp.py under omni.isaac.franka. And the end effector position offset is computed using target position and position of “panda_rightfinger”.

@newuhe How are you determining the difference between the converged end effector position and the target position? RmpFlow and Inverse Kinematics have algorithmically nothing to do with each other, so if they are both bringing the robot the same distance from your target, it sounds like they have …

I print the target position and the converged end effector position and find that there is offset. In the converged initial configuration, it shows end position: (array([-0.02573992, 0.13310829, 0.72076553], dtype=float32), array([ 0.02917537, 0.9953671 , -0.02766433, -0.08733831], dtype=float32…

@newuhe I notice two things from this script See the lines bad_proportional_gains = articulation_controller.get_gains()[0]/50 articulation_controller.set_gains(kps = bad_proportional_gains) These are intentionally setting the PD gains to incorrect values to facilitate the usage of RmpFlow Debug…

@newuhe , were the previous comments helped you to resolve the issue?

Thanks for replying and my problem is solved.

How do you resolve the issue? I am using controller.get_motion_policy().get_end_effector_as_prim().get_world_pose() There is still a good amount of offset.

I do use controller.get_motion_policy().get_end_effector_as_prim().get_world_pose() and the offset is within 0.1mm now.

can you try this position: np.array([0.5, -0.2, 0.28])

End-effector offset of RMP/IK

Omniverse Isaac Sim

arudich November 8, 2022, 6:48pm 26

@nikepupu9
To disable RmpFlow’s self-collision avoidance behavior altogether, you will need to edit the rmpflow_config.yaml file as I described in my last reply. Just make the radii of the body_cylinders and body_collision_controllers very small numbers.

Topic		Replies	Views
Custom Robot Arm not reaching target while using RMP Flow \| Imprecise control of robotic arm Isaac Gym	12	1237	October 5, 2023
RMPflow Motion Generation Isaac Sim	4	657	November 6, 2024
Robot collision sphere moved from robot while using RMPflow Isaac Sim python , custom-robot , robot-pose	8	1595	October 27, 2023
After the RmpFlow exceeds a certain position range, the end following task is not accurate Isaac Sim python , franka-arm	1	266	May 6, 2024
Why get wrong position and orientation using rmpflow controller? Isaac Gym	0	73	December 28, 2024
Lula RMPflow: Collision Spheres Do Not Move With Robot in Tutorial Code Isaac Sim isaacsim , isaac-sim-v5-0-0	4	187	June 10, 2026
Why is it that my robotic arm doesn't work properly with the rmpflow tracking algorithm, what could be the reason? Isaac Sim isaac-sim-v4-5-0	2	140	May 7, 2025
Bug from follow_target_with_rmpflow.py Isaac Sim	1	146	August 10, 2024
Applying action from ArticulationKinematicsSolver always leads to a deviation from the target position for Fetch Robot Isaac Sim	8	808	October 31, 2023
Robot Manipulator does not reach target position RMP Flow Isaac Sim	3	203	December 25, 2024

End-effector offset of RMP/IK

Related topics