Sim2Real with OMY: From Gazebo and IsaacSim Training to Real-World Deployment

:waving_hand: Hello everyone!
In today’s video, we’re excited to share a complete Sim2Real pipeline using the OMY robot — from reinforcement learning in simulation to real-world deployment.

:backhand_index_pointing_right: Introducing OMY
OMY is a compact 6-DoF manipulator with a gripper, tailored for Physical AI and robotics learning. It supports both simulation and real-world operation, and its modular structure makes it ideal for research in imitation learning, reinforcement learning, and Sim2Real transfer.

:hammer_and_wrench: Simulation-to-Reality Workflow
We begin in Isaac Sim, where the OMY robot is trained via reinforcement learning to perform a Reach task. After verifying the policy in Isaac Sim, we move on to Gazebo for Sim2Sim validation. The trained policy is then deployed on a Dynamixel-based dummy robot, where parameter tuning is conducted. Finally, the policy is transferred to the real OMY robot, completing the Sim2Real process.

:folded_hands: Thank you so much for watching!
If you have any questions or ideas, feel free to leave a comment. Your support helps us build better tools for the future of robotics and AI research.

:television:Video

:ledger: Document

:inbox_tray: GitHub Repository

#ROBOTIS #AIWorker #AIManipulator #OMY #OMX #DYNAMIXEL robot #Humanoid ai #OpenSource ROS #PhysicalAI #EmbodiedAI

1 Like