[Iron] [Cartesian Compliance Controller] High stiffness values will result in uncontrolled behaviour when using ros2 control mocked hardware

[firesurfer]

Problem description

Hi, we realized that when setting very high stiffness values (>5000) the cartesian compliance controller ouputs starts jumping around in the weirdest ways. We couldn't reproduce this behaviour on hardware yet. (And also didnt really push for it on real hardware)

Software versions

• Commit: ef2abde
• OS: Ubuntu 22.04
• ROS version: Iron
• Robot: UR16e - mocked hardware interface

To Reproduce

Simply set very high stiffness values when using the mocked hardware interface -> No valid FTS feedback!

Expected behavior

With mocked hardware -> Controller behaves like a position controller
With real hardware -> We often try to be compliant in only one direction

Additional context

With mocked hardware there is no valid Force Torque sensor feedback

[stefanscherzinger]

Thanks for reporting this! I'll need to reproduce this and check what happens.

---

Edit: @firesurfer if this is still relevant, could you post your corresponding controller parameters here? I might spot some unusual combination of parameters to get a starting idea for debugging.