[Question] Motion Control Anomaly with Realman Manipulator in Isaac Sim

[shensikang]

We are currently controlling a Realman manipulator (with joint configuration similar to Franka Panda) in the Isaac Lab 2.1 simulation environment and encountering an issue with motion control.

​​Expected Behavior:​​
When sending control commands, we intend for the end-effector / terminal link (Link 7) to perform ​​pure translational motion​​ along the positive X-axis of the base_link coordinate frame (i.e., the globally defined "forward" direction).

​​Observed Behavior:​​
The actual movement in simulation severely deviates from expectations. The end-effector (Link 7) fails to execute the desired forward translational motion.

Instead, the motion appears to be primarily driven by ​​Link 6​​ (the direct parent link of Link 7). Specifically:

• Link 7 shows minimal forward displacement
• The manipulator exhibits dominant motion characteristics of ​​Joint 6​​ (connecting Link 6)
• Manifestations include:
  • Link 7 retracting backward or moving laterally
  • Unplanned rotational movements (yaw/rotation around Joint 6)
  • Failure to translate along its local forward direction

2025.07.17.10.56.59.webm

[RandomOakForest]

Thanks for posting this. If you move one joint at a time, do all go in the direction you expect? This may help debug reference system direction. It's possible that the Panda and Realman have similar joint configurations but some with opposite orientations. Also it would help if you post an example that works with the Franka Panda so we understand what you are trying to do.

[shensikang]

Thanks for posting this. If you move one joint at a time, do all go in the direction you expect? This may help debug reference system direction. It's possible that the Panda and Realman have similar joint configurations but some with opposite orientations. Also it would help if you post an example that works with the Franka Panda so we understand what you are trying to do.

I’m using inverse kinematics (IK) to control the robotic arm’s end effector. While controlling individual joints works fine. when moving the entire arm along the x-axis to a certain point, it suddenly accelerates and becomes unstable. Reducing the control command mitigates this, but the issue persists at specific positions.

[shensikang]

the robot might be hitting a ​​singularity point​​ . Do you have any good solutions or suggestions for this?

[RandomOakForest]

Thank you for following up. Here are a few pointers:

• Use "dls" (Damped Least Squares) IK method with tuned damping for improved stability at singularities.
• Monitor Jacobian condition number/minimum singular value for early detection.
• Avoid workspace boundaries and singular pose targets.
• Rely on Isaac Sim 5.0 and Isaac Lab 2.2 controller APIs and documentation to configure IK solvers robustly. See Using Isaac Lab with Isaac Sim 5.0 in a Conda Environment #3021 to update.

I'll move this post to our Discussions for follow up.