Downloads · 30 days
0
Mattewkang/Abstraction_experiment-arm-models
Abstraction_experiment-arm-models is a robotics model from Mattewkang. Use it for the robotics task on the model card, and read the license before you ship it in a product.
Weights and planner data used by the g1armabs package of Abstractionexperiment (Unitree G1 vision-guided grasping, elevator-button press and berry picking, ROS Noetic).
Downloads · 30 days
0
Access
Public
Updated Sep 23, 2026
Repo size
29.7 GB
Likes
0
Public
Click a slice to open those files.
.uint829.7 GB · 100%
From the Hugging Face model README
Weights and planner data used by the g1_arm_abs package of
Abstraction_experiment
(Unitree G1 vision-guided grasping, elevator-button press and berry picking, ROS Noetic).
| Directory | Contents | Used by |
|---|---|---|
grasp_bottle/{left,right}/ | MLP inverse-kinematics model (best_model.pth) + input scaler (x_scaler.pkl) per arm | grasp_bottle, *_grasp_from_point services |
elevator_button/{left,right}/ | MLP IK model + scaler + config.json per arm | press_elevator_button |
grasp_berry/{left,right}/ | MLP IK model + scaler per arm | pick_berry |
abstraction_map/ | Voxel A* planner data: per-arm reachability bitmaps (voxel_config_bitmap_{left,right}.uint8, ~14 GB each), feasibility masks, sparse voxel maps, phase-2 metadata and logs | Obstacle-aware A* on every planned motion |
The two voxel_config_bitmap_*.uint8 files total about 28 GB; make sure you have
the disk space before fetching.
The project pins this repo's commit and every file's sha256 in
g1_arm_abs/config/model_weights.yaml; rosrun g1_arm_abs fetch_weights.py
downloads and verifies them into g1_arm_abs/models/ preserving this directory
layout. Manual download:
hf download Mattewkang/Abstraction_experiment-arm-models --local-dir g1_arm_abs/models