Downloads · 30 days
0
oku006/bussing-pi05-droid
bussing-pi05-droid is a machine learning model from oku006. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Deploy via branch pi05-joint-velocity of sombit888/robotevals — ./setup.sh + pi05jvelrobot.yaml. The branch pins this repo's revision and carries the required latched policy; do not hand-wire this checkpoint.
Downloads · 30 days
0
Access
Public
Updated Aug 25, 2026
Repo size
49.7 GB
Likes
1
Public
Click a slice to open those files.
Other12.4 GB · 100%
From the Hugging Face model README
Deploy via branch
pi05-joint-velocityofsombit888/robot_evals—./setup.sh+pi05_jvel_robot.yaml. The branch pins this repo's revision and carries the required latched policy; do not hand-wire this checkpoint.
π0.5 fine-tuned from pi05_droid on the FR3 bussing task.
Action space: 7 joint velocities (rad/s) + gripper position (8-d).
Run at 5 Hz (min_period: 0.2) — the training tick and execution tick are
both 0.2 s. flip_gripper: True — trained on the DROID gripper convention
(1-g of the robot's raw values); the runner flips state in and action out.
The previous revision baked a ×3 velocity factor into the norm stats after
training — each executed tick covered ~3 frames of demo path while the model's
chunk clock advanced one. This revision trains the speed-up natively:
velocity targets are derived across a 3-frame stride, (q[t+3]-q[t])/0.2,
so one 0.2 s tick covers 3 frames of path by construction. Timeline exact,
3× demo speed, no post-hoc scaling anywhere (VELOCITY_SCALE.json = 1.0).
Training data: idle runs cut with episodes split at every cut (no chunk spans a cut), grasp/release events explicitly protected from the idle cutter (without that rescue 107 of 292 gripper transitions sat inside cut regions), 3 phase-offset sub-episodes so every original frame is used.
| check | result |
|---|---|
| cold-scene gate (moves + aims like the operator) | 7/10, mean |v| 0.055 rad/s (frozen predecessor: 0.008) |
| phase cosines vs demonstrator | approach 0.55 · grasp 0.57 · transport 0.55 · release 0.62 |
| gripper transition timing | median 1 tick lead/lag, 17/95 missed |
| FR3 joint limits after integrating the 8 executed ticks | 0 violations |
| stride indexing, all 16 chunk ticks | exact |
| post-hoc scaling | none |
Register pi05_droid_bussing_5hz (see train_config_snippet.py), norm stats
ship in assets/bussing_droid_5hz_s3/. Action horizon 16, first 8 executed.
Instruction: put the vegetables in the bowl.