Downloads · 30 days
0
EdwardoSunny/pi05_xarm_cube_binmarker9_lora
pi05_xarm_cube_binmarker9_lora is a robotics model from EdwardoSunny. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for openpi. The card lists the license as apache-2.0.
pi0.5 LoRA fine-tune for the xArm cube task ("pick up the cube"), trained on frames with the binmarker9 tactile overlay burned in.
Downloads · 30 days
0
Access
Public
Updated Sep 6, 2026
Repo size
6.3 GB
Likes
0
Public
Click a slice to open those files.
Other6.3 GB · 100%
From the Hugging Face model README
binmarker9 (9 arrows/finger, contact bit only)pi0.5 LoRA fine-tune for the xArm cube task ("pick up the cube"), trained on frames
with the binmarker9 tactile overlay burned in.
Only the per-finger binary contact bit. Both force direction and force magnitude are omitted:
| Signal | Present? |
|---|---|
| Binary contact bit | YES |
| Force direction | NO - fixed to the sensor-frame pad normal |
| Force magnitude | NO - fixed length (0.12 m) |
9 arrows per finger (one at each of the 9 forward-kinematics-projected sensor pad cells), drawn iff that finger's contact state is ON (max-cell metric, hysteresis ON >= 0.25 / OFF < 0.18, 2-on/3-off debounce, state reset per episode).
Verified tactile-independence: 0 max pixel change across 300 randomized force readings at a fixed pose - the pixels provably carry no force information.
| Task | 1 arrow | 9 arrows |
|---|---|---|
| cube | binmarker | binmarker9 |
Hyperparameters are identical across both arms (LoRA gemma_2b_lora +
gemma_300m_lora, batch size 8, cosine schedule warmup 500 / peak 1e-4 /
decay 20k to 1e-5, AdamW grad-clip 1.0, no EMA, 20k max steps with early
stopping), so differences are attributable to the overlay alone.
Trained 7900 steps.
Orbax checkpoint (params/) plus assets/ norm stats - inference-ready.
train_state/ (optimizer state) is intentionally excluded.
python examples/xarm/inference/run_xarm_inference.py \
--checkpoint <local-path> \
--overlay binmarker9_cube
The overlay must be redrawn at deployment exactly as in training, or the
policy is out of distribution. The inference renderer (FixedMarkerOverlay,
variant binmarker9) is verified byte-exact against the training renderer:
0 max pixel error over 3,566 camera images across all 4 tasks (cube 450,
tube 760, charger 1,122, dishwasher 1,234), 0 contact-state mismatches. The --overlay flag requires the per-task suffix
because the per-finger p99 scales differ up to ~6x across tasks.
Observation/action convention: state 8-dim
[ee_pos_m(3), ee_axis_angle_rad(3), grasp, grasp], action 7-dim
[dxyz_m, d_axis_angle_rad, grasp_pm1], 224x224 RGB uint8, 10 Hz, action
chunk 10 (consume 5).