rskill-gr00t-n17-b1k-turning-on-radio
OpenRAL rSkill — the official 2026 BEHAVIOR-1K GR00T N1.7
turning_on_radio checkpoint for the simulated Galaxea R1 Pro.
This package contains the OpenRAL manifest and adapter configuration only. The
organizer checkpoint is downloaded separately from the
BEHAVIOR baseline page;
weights are not copied into this repository.
Preview
The official task demonstration is available in the
BEHAVIOR challenge gallery.
No local rollout image is shipped until the checkpoint has been reproduced on
this host.
What this skill does
The policy navigates the R1 Pro to a household radio and manipulates its controls
to turn it on. It is task-specific to turning_on_radio.
| Field | Value |
|---|
| Actions | rotate, push |
| Objects | radio, dial, button |
| Scenes | household living room |
| Embodiment | BEHAVIOR R1 Pro (r1pro) |
How it works
The adapter runs the pinned wensi-ai/Isaac-GR00T behavior branch in an
isolated Python 3.10 sidecar. The sidecar uses the upstream Gr00tPolicy and
B1KPolicyWrapper unchanged, including temporal ensembling and the official
R1Pro modality split. Two 8 GB-host memory measures are applied at load time:
whole-model NF4 quantization of large linears, and replacing the Qwen3-VL
lm_head with Identity (the wrapper consumes only hidden states, so the
full-vocab logits projection is dead weight).
Observation -> action contract
| Direction | Key | Shape | Notes |
|---|
| in | head RGB | HWC uint8 | Official ZED head camera |
| in | left wrist RGB | HWC uint8 | Left RealSense |
| in | right wrist RGB | HWC uint8 | Right RealSense |
| in | proprioception | (61,) float32 | Official PROPRIOCEPTION_INDICES["R1Pro"] order |
| out | action | (23,) float32 | Base velocity 3 + torso 4 + arms 7+7 + grippers 1+1 |
The organizer wrapper converts the 61-D vector into base, torso, arm, and
gripper state groups. It emits a 16-step GR00T action horizon and returns one
temporally-ensembled 23-D action per evaluator step.
Upstream model / training
Supported robots
| Robot | Embodiment tag | Status | Notes |
|---|
| Simulated Galaxea R1 Pro | r1pro | evaluator + deploy sim | Official BEHAVIOR observation/action contract |
openral deploy sim uses robots/r1pro/robot.yaml, publishes the simulator's
native 61-D policy state through WorldState, validates every typed action slot
through the safety kernel, then atomically commits all six slots as one 23-D
OmniGibson step.
Sensors required
| Key | Modality | Min resolution | Format |
|---|
observation.images.head | RGB | 224 x 224 | HWC uint8 |
observation.images.left_wrist | RGB | 224 x 224 | HWC uint8 |
observation.images.right_wrist | RGB | 224 x 224 | HWC uint8 |
observation.state | proprioception | (61,) | float32 |
Manifest summary
| Field | Value |
|---|
name | OpenRAL/rskill-gr00t_n17-r1pro-turning_on_radio-bf16 |
version | 0.1.0 |
license | unknown |
role | s1 |
model_family | gr00t |
runtime | external Python 3.10 Isaac-GR00T sidecar, whole-model NF4 |
weights_uri | local://checkpoints/behavior-groot-turning-on-radio |
chunk_size | 16 |
state_contract.dim / action_contract.dim | 61 / 23 |
latency_budget.per_chunk_ms | 1500 |
Quick start
1git clone https://github.com/wensi-ai/Isaac-GR00T \
2 ~/.cache/openral/behavior-groot/source
3git -C ~/.cache/openral/behavior-groot/source checkout \
4 ace36d935b376fbf25cd56371e23877b95407c40
5cd ~/.cache/openral/behavior-groot/source
6uv sync --frozen --python 3.10
7
8export OPENRAL_BEHAVIOR_GROOT_SIDECAR_PYTHON="$PWD/.venv/bin/python"
9export OPENRAL_BEHAVIOR_GROOT_CHECKPOINT=/absolute/path/to/checkpoint
10
11cd /path/to/openral
12just sync --group behavior-groot
13openral behavior serve \
14 --rskill rskills/gr00t-n17-b1k-turning-on-radio \
15 --task turning_on_radio
Full deploy graph:
1openral deploy sim \
2 --config scenes/deploy/behavior_r1pro.yaml \
3 --initial-task "turn on the radio"
In the BEHAVIOR environment:
1python -m omnigibson.eval.eval \
2 --task-name turning_on_radio \
3 --host 127.0.0.1 --port 8000 \
4 --instance-indices 0 --num-rollouts 1 \
5 --output-dir outputs/openral --write-video
Reproduction
The command above is the canonical one-rollout reproduction. Use public
instances 0-9 for challenge reporting and keep the evaluator-generated JSON
and videos unmodified.
Evaluation
No OpenRAL-generated score is shipped. The full official-evaluator loop has
been reproduced locally on an 8 GB RTX 4070 Laptop GPU (NF4 sidecar 2.77 GiB
inference peak alongside OmniGibson, ~0.8-1.3 steps/s): public instance 0
(ID 301), 3225 steps to timeout, success=false, q_score 0.0. Whether the
zero q_score reflects NF4 degradation or the checkpoint's zero-shot behavior
on this instance has not been isolated; no success-rate claim is made.
License
The OpenRAL adapter, manifest, and documentation are Apache-2.0. The
organizer-provided fine-tuned checkpoint is marked license: unknown because
the Google Drive artifact has no separately published license file. The base
GR00T N1.7 model uses the NVIDIA Open Model License, but this package does not
silently assume that the fine-tune inherits identical terms.
See also
