Datasets:

---

lerobot

/

libero_spatial_image

like 3

Follow

LeRobot 3.25k

Hi,

Thanks for the great work!

I notice that the Libero data provided by OpenVLA (https://huggingface.co/datasets/openvla/modified_libero_rlds/blob/main/libero_spatial_no_noops/1.0.0/features.json ) contains the observation.state which is composed of "Robot EEF state (6D pose, 2D gripper)." While in the Lerobot description, the state is composed of x-y-z-rx-ry-rz-rw-gripper.

Could you help me double check that the Lerobot state contains only 1 dimension for gripper, 7 dimensions for pose? And 1 for gripper open, -1 for gripper close'?

Best regards,
Xianjie