In my Mujoco simulation, the joint angle motion never tracks my target joint settings well.

[xiaozhu1]

All my joints are set to position control mode, and I am using version 4.10. I have also tried other physics engines, but with those the simulation results in very unstable oscillations. Regarding my joint angle target tracking, the performance is especially poor for the seventh joint, which is a prismatic joint. I have also tried tuning the PID parameters by increasing P and D, but this had little effect. At this point, I am not sure where the problem lies. Would switching to another version improve this, or is there something wrong with my current settings? I have also uploaded my scene, and I would greatly appreciate your help.https://github.com/wobushixiaozhu/coppe ... it_arm.git

[coppelia]

Hello,

your scene show an immobile robot: positions appear to be holding in a stable fashion. You are probably using the remote API to move the robot around. I also noticed that you are probably using the legacy remote API, which we do not support anymore. Make sure to switch to the ZeroMQ remote API instead.

Also, you might have to adjust the armature parameter of the various joints of your robot for better stability, and/or increase the max. velocity and/or max. torque for those joints.

Finally, make sure that there are no robot self-collisions occuring. Also, object board0 is not convex and might not be simulated in an optimal manner.

Cheers

[xiaozhu1]

Thank you so much for your help. I’ve implemented my functionality via the ZMQ API. I now have a few new questions: I want my end-effector gripper to interact with the ring in the environment. How should I set this up—should I use collision detection or some other approach? Thanks again for your help. Lastly, here is the link to my new scene.
https://github.com/wobushixiaozhu/NEW_DVRK_ARM.git