Views
No views yet
sl_checkpoint_199.pth: Trained policy weights (ViLGP3D fusion network)checkpoint-rs.tar: GraspNet checkpoint for grasp candidate generation1from lerobot_policy_a2 import A2Policy
2
3# Load pretrained model
4policy = A2Policy.from_pretrained("dgrachev/a2_pretrained")
5
6# Use for grasp prediction
7action, info = policy.predict_grasp(
8 color_images={"front": rgb_image},
9 depth_images={"front": depth_image},
10 point_cloud=point_cloud,
11 lang_goal="grasp a round object"
12)1@misc{a2_policy,
2 author = {Denis Grachev},
3 title = {A2 Policy: CLIP-based 6-DOF Grasp and Place Policy},
4 year = {2025},
5 publisher = {HuggingFace},
6 url = {https://huggingface.co/dgrachev/a2_pretrained}
7}