POLICY · CODING · #2375
Train in simulation and deploy walking policies to Pollen Robotics' Microduck with NVIDIA Isaac Lab 3.0
A newly published tutorial demonstrates how to train bipedal walking policies for the open-source Microduck robot using NVIDIA Isaac Lab 3.0 and Newton actuator models, export the policy as ONNX, and deploy it on the physical robot. The workflow uses GPU-parallel simulation, domain randomization, Newton/BAM actuator and backlash modeling, and automatic RSL-RL playback exports so models can be shared via the Hugging Face Hub or tested in Microduck Arena.
KEY POINTS
- A newly published tutorial demonstrates how to train bipedal walking policies for the open-source Microduck robot using NVIDIA Isaac Lab 3.0 and Newton actuator models, export the policy as ONNX, and deploy it on the physical robot.
- The workflow uses GPU-parallel simulation, domain randomization, Newton/BAM actuator and backlash modeling, and automatic RSL-RL playback exports so models can be shared via the Hugging Face Hub or tested in Microduck Arena.
- The integration lowers barriers for sim-to-real bipedal locomotion by combining Isaac Lab 3.0's GPU-parallel training and Newton actuator modeling with an affordable open-source robot and ONNX deployment.
WHY IT MATTERS
The integration lowers barriers for sim-to-real bipedal locomotion by combining Isaac Lab 3.0's GPU-parallel training and Newton actuator modeling with an affordable open-source robot and ONNX deployment.