GR00T E2E Reference Workflow#
REFERENCE WORKFLOW
Build robot-learning systems end to end
See how NVIDIA robotics and AI products work together in an end-to-end system. Follow complete workflow paths that connect data collection to trained policies in simulation and on physical robots.
SIM→DATA→POLICY→EVAL
Featured Workflow Paths#
AVAILABLE
Unitree G1 · NVIDIA Isaac Lab · GR00T N1.7
Record demonstrations, convert the data, fine-tune a policy, and evaluate it in the same scene.
Open workflow path →
AVAILABLE
Unitree G1 · Isaac ROS · Jetson AGX Thor · GR00T N1.7
Collect physical demonstrations, fine-tune and export a policy, then deploy and evaluate it on the Unitree G1.
Open workflow path →
AVAILABLE
Unitree H2 Plus · NVIDIA Isaac Lab-Arena · AutoData · GR00T N1.7
Record bimanual demonstrations, optionally expand them with AutoData, then convert the data, train a policy, and evaluate it in simulation. Policy training is validated. Learned-policy evaluation is implemented but not yet validated.
Open workflow path →
WORKFLOW AUTHORS
Use the existing workflow requirements to add another environment, robot configuration, or policy path and register its documentation in the site navigation.
Build a Workflow Path →