cagataydev2026cc-by-4.0-generated-with-nvidia-isaac
Unitree G1 humanoid rough-terrain locomotion (Isaac Lab, recorded with strands-robots)
A PPO policy trained in NVIDIA Isaac Lab walks the 37-DoF Unitree G1 humanoid over procedurally generated rough terrain while tracking commanded base velocity. Rollouts were recorded as a LeRobot v3 dataset with 12 episodes of 6000 frames at 50 fps.
Downloads76
Episodes12
Hours0.0333
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- Unitree G1humanoid
- Action Space
- 37-DoF: joint_pos.left_hip_pitch_joint, joint_pos.right_hip_pitch_joint, joint_pos.torso_joint, joint_pos.left_hip_roll_joint, joint_pos.right_hip_roll_joint, joint_pos.left_shoulder_pitch_joint, joint_pos.right_shoulder_pitch_joint, joint_pos.left_hip_yaw_joint, joint_pos.right_hip_yaw_joint, joint_pos.left_shoulder_roll_joint, joint_pos.right_shoulder_roll_joint, joint_pos.left_knee_joint
- Environment
- simulation
- Task Types
- locomotionrough-terrain
- Episodes
- 12
- Total Hours
- 0.0333
- Data Format
- LeRobot
- License
- cc-by-4.0-generated-with-nvidia-isaac
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack