We have released our technical guide and demo on VR teleoperation for the ROBOTIS AI Worker in NVIDIA Isaac Sim.
To support Physical AI development and imitation learning, we built a ROS 2-based DDS bridge. This allows developers to test mobile manipulation tasks and collect imitation learning data in simulation without needing the physical hardware immediately.
To ensure the virtual environment behaves like the real world, we focused on crucial physical tuning:
- DDS Bridge: Maps joint trajectory and cmd vel commands from the ROS 2 control node (Cyclo Motion Controller) in real-time.
- High-Friction Fingertips: Mimics real rubber properties for secure and stable grasping.
- Collision Filtering: Eliminates self-collision interference between fingers and the swerve drive joint for smooth, precise control.
It includes:
- VR-based manipulation control
- Keyboard-based mobile base control
- Object interaction in Isaac Sim
- Virtual task environments such as kitchen/table and warehouse scenes
Check out our detailed architecture and Docker setup to run this yourself.
Happy researching! 
Video: https://youtu.be/4D8if2lZY7o
Code: https://github.com/ROBOTIS-GIT/robotis_lab
Docs: AI Worker SH5 Isaac Sim VR Teleoperation | ROBOTIS Docs