[RANGE ] 安全範囲 (学習データ ±10°) は /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes から (--model deep1)
[FRAME ] 写り方の確かめの基準画像は /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes の最初のエピソードから (位置合わせと同じ trim2. 依頼 137)
[TILT  ] 上カメラの傾きのお手本 (各エピソードの 1 枚目の中央値) は /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes から (安全範囲と同じ. 差だけ画像を回して AI に渡す. 依頼 144)
[GRIP  ] 爪の閉じる値は 25 (--grip-value). 締め直し r はこの値−3 とこの値を交互に送る (依頼 152)
[LAB   ] 2026-11-11 16:01:03  GPU=NVIDIA A40  episode=run_0001_20261019_194818  log=/Volumes/KIOXIA/r2r2r-data/lab/Log30  
[MODEL ] --model deep1: π0.5 base から deep_even10_v1 で新しく学習した 2999 (依頼 133. decay_steps 5000)
         PixelstoMotors/pi05-mecharm-deep-even10-lora-v1 → /workspace/ckpt/deep_v1_2999  設定 pi05_mecharm_deep_even10_lora  assets local/mecharm_redblock_box_deep_even10_v1

=== 1. Pod を作る
[POD   ] 記録の Pod (id t3fnpcsccc95vo) は API で見ると残っていない. 新しく作ります
         古い記録は /Volumes/KIOXIA/r2r2r-data/lab/Log30/pod/lab_pod_20261111_160104.json に残しました
[POD   ] 作った: id 1zx0vdv9xasenp  GPU=NVIDIA A40  1時間 0.59 ドル  (記録 /Volumes/KIOXIA/r2r2r-data/lab/Log30/pod/lab_pod.json)
    (4 秒)

=== 2. Pod が立ち上がるまで待つ
[POD   ] SSH の IP とポートを API から受け取った (<POD_IP>:<SSH_PORT>)
[POD   ] 入れた: 6f893fdedff0
NVIDIA A40, 46068 MiB
    (40 秒)

=== 3a. スクリプトを送る
    (122 秒)

=== 3b. huggingface_hub を入れ, 鍵を渡す
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning.
Hint: The `hf-cli` skill is not installed. Run `hf skills add -g` to teach your AI agents how to use the `hf` CLI.
✓ Logged in
  user: PixelstoMotors
    (14 秒)

=== 3c. モデルを取る (hf download PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)

Fetching 34 files:   0%|          | 0/34 [00:00<?, ?it/s]
Fetching 34 files:   3%|▎         | 1/34 [00:00<00:09,  3.49it/s]
Fetching 34 files:  26%|██▋       | 9/34 [00:00<00:01, 18.45it/s]
Fetching 34 files:  32%|███▏      | 11/34 [00:00<00:01, 18.76it/s]
Fetching 34 files:  38%|███▊      | 13/34 [00:01<00:02,  8.99it/s]
Fetching 34 files:  44%|████▍     | 15/34 [00:03<00:07,  2.38it/s]
Fetching 34 files:  47%|████▋     | 16/34 [00:05<00:10,  1.65it/s]
Fetching 34 files:  50%|█████     | 17/34 [00:05<00:09,  1.84it/s]
Fetching 34 files:  56%|█████▌    | 19/34 [00:05<00:06,  2.47it/s]
Fetching 34 files:  59%|█████▉    | 20/34 [00:06<00:05,  2.63it/s]
Fetching 34 files:  62%|██████▏   | 21/34 [00:06<00:05,  2.30it/s]
Fetching 34 files:  65%|██████▍   | 22/34 [00:07<00:04,  2.52it/s]
Fetching 34 files:  68%|██████▊   | 23/34 [00:07<00:03,  3.00it/s]
Fetching 34 files:  71%|███████   | 24/34 [00:07<00:03,  2.62it/s]
Fetching 34 files:  76%|███████▋  | 26/34 [00:07<00:02,  3.58it/s]
Fetching 34 files:  82%|████████▏ | 28/34 [00:08<00:01,  5.22it/s]
Fetching 34 files:  91%|█████████ | 31/34 [00:08<00:00,  7.66it/s]
Fetching 34 files:  97%|█████████▋| 33/34 [00:09<00:00,  4.77it/s]
Fetching 34 files: 100%|██████████| 34/34 [00:09<00:00,  4.88it/s]
Fetching 34 files: 100%|██████████| 34/34 [00:09<00:00,  3.69it/s]
✓ Downloaded
  path: /workspace/ckpt/deep_v1_2999
6.0G	/workspace/ckpt/deep_v1_2999
    (14 秒)

=== 3d. openpi を取る (215abfb)
215abfb docs(droid): fix config search instruction (#1023)
    (22 秒)

=== 3e. uv sync
downloading uv 0.13.7 x86_64-unknown-linux-gnu
WARN: The following commands are shadowed by other commands in your PATH: uv uvx
Using CPython 3.11.13 interpreter at: /usr/bin/python3.11
Creating virtual environment at: .venv
Resolved 279 packages in 4ms
   Building openpi-client @ file:///workspace/openpi/packages/openpi-client
   Building openpi @ file:///workspace/openpi
   Updating https://github.com/huggingface/lerobot (0cf864870cf29f4738d3ade893e6fd13fbd7cdb5)
Downloading numpy (17.4MiB)
Downloading nvidia-cublas-cu12 (374.9MiB)
Downloading pyarrow (40.3MiB)
Downloading sympy (6.0MiB)
Downloading scipy (35.9MiB)
Downloading torchvision (7.1MiB)
Downloading cmake (26.6MiB)
Downloading av (34.7MiB)
Downloading nvidia-cusparse-cu12 (206.5MiB)
Downloading h5py (4.3MiB)
Downloading nvidia-cusparselt-cu12 (149.5MiB)
Downloading opencv-python (60.1MiB)
Downloading triton (148.5MiB)
Downloading rerun-sdk (68.1MiB)
Downloading diffusers (3.4MiB)
Downloading numcodecs (8.4MiB)
Downloading nvidia-nvjitlink-cu12 (18.8MiB)
Downloading nvidia-cuda-nvrtc-cu12 (22.6MiB)
Downloading nvidia-cuda-nvcc-cu12 (38.7MiB)
Downloading nvidia-cusolver-cu12 (150.9MiB)
Downloading matplotlib (8.2MiB)
Downloading mujoco (4.4MiB)
Downloading imageio-ffmpeg (28.1MiB)
Downloading nvidia-cudnn-cu12 (544.5MiB)
Downloading transformers (10.3MiB)
Downloading fonttools (4.7MiB)
Downloading numba (3.7MiB)
Downloading hf-xet (4.9MiB)
Downloading polars (34.6MiB)
Downloading nvidia-cufft-cu12 (190.9MiB)
Downloading dm-control (37.5MiB)
Downloading lxml (4.7MiB)
Downloading torch (783.1MiB)
Downloading jaxlib (100.2MiB)
Downloading pyopengl (3.0MiB)
Downloading jax-cuda12-pjrt (99.8MiB)
Downloading virtualenv (5.8MiB)
Downloading nvidia-nccl-cu12 (192.0MiB)
Downloading tensorstore (17.5MiB)
Downloading pandas (12.5MiB)
Downloading nvidia-curand-cu12 (53.7MiB)
Downloading ruff (11.0MiB)
Downloading labmaze (4.7MiB)
Downloading wandb (20.4MiB)
Downloading opencv-python-headless (47.7MiB)
Downloading nvidia-cuda-cupti-cu12 (8.5MiB)
Downloading pillow (4.4MiB)
Downloading hf-transfer (3.4MiB)
Downloading jax-cuda12-plugin (15.9MiB)
Downloading llvmlite (40.4MiB)
 Downloaded pyopengl
Downloading debugpy (3.0MiB)
 Downloaded hf-transfer
 Downloaded diffusers
Downloading tokenizers (2.9MiB)
 Downloaded numba
Downloading jax (2.3MiB)
Downloading widgetsnbextension (2.1MiB)
 Downloaded h5py
Downloading ml-dtypes (2.1MiB)
 Downloaded pillow
 Downloaded mujoco
Downloading networkx (1.9MiB)
 Downloaded labmaze
Downloading pydantic-core (1.9MiB)
 Downloaded fonttools
Downloading aiohttp (1.6MiB)
 Downloaded lxml
Downloading jedi (1.5MiB)
 Downloaded hf-xet
Downloading kiwisolver (1.4MiB)
Downloading torchcodec (1.3MiB)
 Downloaded virtualenv
 Downloaded widgetsnbextension
Downloading sentencepiece (1.2MiB)
 Downloaded sympy
 Downloaded jax
Downloading pygments (1.2MiB)
Downloading setuptools (1.1MiB)
 Downloaded debugpy
Downloading nvidia-cufile-cu12 (1.1MiB)
 Downloaded ml-dtypes
Downloading pymunk (1.0MiB)
 Downloaded torchcodec
 Downloaded kiwisolver
 Downloaded tokenizers
 Downloaded networkx
 Downloaded aiohttp
 Downloaded pydantic-core
 Downloaded jedi
 Downloaded torchvision
 Downloaded sentencepiece
 Downloaded setuptools
 Downloaded pygments
 Downloaded nvidia-cufile-cu12
 Downloaded pymunk
 Downloaded matplotlib
 Downloaded numcodecs
 Downloaded nvidia-cuda-cupti-cu12
   Building evdev==1.9.2
 Downloaded transformers
   Building antlr4-python3-runtime==4.9.3
 Downloaded ruff
 Downloaded pandas
 Downloaded jax-cuda12-plugin
 Downloaded tensorstore
 Downloaded numpy
 Downloaded nvidia-nvjitlink-cu12
      Built openpi @ file:///workspace/openpi
      Built openpi-client @ file:///workspace/openpi/packages/openpi-client
 Downloaded nvidia-cuda-nvrtc-cu12
 Downloaded wandb
 Downloaded imageio-ffmpeg
      Built antlr4-python3-runtime==4.9.3
 Downloaded av
 Downloaded polars
 Downloaded scipy
 Downloaded dm-control
 Downloaded cmake
 Downloaded nvidia-cuda-nvcc-cu12
 Downloaded llvmlite
 Downloaded pyarrow
 Downloaded opencv-python-headless
 Downloaded nvidia-curand-cu12
 Downloaded opencv-python
 Downloaded rerun-sdk
      Built evdev==1.9.2
 Downloaded jax-cuda12-pjrt
 Downloaded jaxlib
    Updated https://github.com/huggingface/lerobot (0cf864870cf29f4738d3ade893e6fd13fbd7cdb5)
   Building lerobot @ git+https://github.com/huggingface/lerobot@0cf864870cf29f4738d3ade893e6fd13fbd7cdb5
 Downloaded triton
 Downloaded nvidia-cusparselt-cu12
 Downloaded nvidia-cusolver-cu12
 Downloaded nvidia-nccl-cu12
 Downloaded nvidia-cufft-cu12
 Downloaded nvidia-cusparse-cu12
      Built lerobot @ git+https://github.com/huggingface/lerobot@0cf864870cf29f4738d3ade893e6fd13fbd7cdb5
 Downloaded nvidia-cublas-cu12
 Downloaded nvidia-cudnn-cu12
 Downloaded torch
Prepared 240 packages in 13.64s
Installed 240 packages in 1.90s
 + absl-py==2.3.0
 + aiohappyeyeballs==2.6.1
 + aiohttp==3.12.4
 + aiosignal==1.3.2
 + annotated-types==0.7.0
 + antlr4-python3-runtime==4.9.3
 + asttokens==3.0.0
 + attrs==25.3.0
 + augmax==0.4.1
 + av==17.0.0
 + beartype==0.19.0
 + beautifulsoup4==4.13.4
 + blinker==1.9.0
 + cachetools==5.5.2
 + certifi==2025.4.26
 + cffi==1.17.1
 + cfgv==3.4.0
 + charset-normalizer==3.4.2
 + chex==0.1.90
 + click==8.2.1
 + cloudpickle==3.1.1
 + cmake==4.0.2
 + comm==0.2.2
 + contourpy==1.3.2
 + crc32c==2.7.1
 + cycler==0.12.1
 + datasets==3.6.0
 + debugpy==1.8.14
 + decorator==5.2.1
 + deepdiff==8.5.0
 + diffusers==0.33.1
 + dill==0.3.8
 + distlib==0.3.9
 + dm-control==1.0.14
 + dm-env==1.6
 + dm-tree==0.1.9
 + docker-pycreds==0.4.0
 + docstring-parser==0.16
 + donfig==0.8.1.post1
 + draccus==0.10.0
 + einops==0.8.1
 + equinox==0.12.2
 + etils==1.12.2
 + evdev==1.9.2
 + executing==2.2.0
 + farama-notifications==0.0.4
 + filelock==3.18.0
 + flask==3.1.1
 + flatbuffers==25.2.10
 + flax==0.10.2
 + fonttools==4.58.1
 + frozenlist==1.6.0
 + fsspec==2025.3.0
 + gcsfs==2025.3.0
 + gdown==5.2.0
 + gitdb==4.0.12
 + gitpython==3.1.44
 + glfw==2.9.0
 + google-api-core==2.24.2
 + google-auth==2.40.2
 + google-auth-oauthlib==1.2.2
 + google-cloud-core==2.4.3
 + google-cloud-storage==3.1.0
 + google-crc32c==1.7.1
 + google-resumable-media==2.7.2
 + googleapis-common-protos==1.70.0
 + gym-aloha==0.1.1
 + gymnasium==0.29.1
 + h5py==3.13.0
 + hf-transfer==0.1.9
 + hf-xet==1.1.2
 + huggingface-hub==0.32.3
 + humanize==4.12.3
 + identify==2.6.12
 + idna==3.10
 + imageio==2.37.0
 + imageio-ffmpeg==0.6.0
 + importlib-metadata==8.7.0
 + importlib-resources==6.5.2
 + iniconfig==2.1.0
 + inquirerpy==0.3.4
 + ipykernel==6.29.5
 + ipython==9.2.0
 + ipython-pygments-lexers==1.1.1
 + ipywidgets==8.1.7
 + itsdangerous==2.2.0
 + jax==0.5.3
 + jax-cuda12-pjrt==0.5.3
 + jax-cuda12-plugin==0.5.3
 + jaxlib==0.5.3
 + jaxtyping==0.2.36
 + jedi==0.19.2
 + jinja2==3.1.6
 + jsonlines==4.0.0
 + jupyter-client==8.6.3
 + jupyter-core==5.8.1
 + jupyterlab-widgets==3.0.15
 + kiwisolver==1.4.8
 + labmaze==1.0.6
 + lerobot==0.1.0 (from git+https://github.com/huggingface/lerobot@0cf864870cf29f4738d3ade893e6fd13fbd7cdb5)
 + llvmlite==0.44.0
 + lxml==5.4.0
 + markdown-it-py==3.0.0
 + markupsafe==3.0.2
 + matplotlib==3.10.3
 + matplotlib-inline==0.1.7
 + mdurl==0.1.2
 + mergedeep==1.3.4
 + ml-collections==1.0.0
 + ml-dtypes==0.4.1
 + mpmath==1.3.0
 + msgpack==1.1.0
 + mujoco==2.3.7
 + multidict==6.4.4
 + multiprocess==0.70.16
 + mypy-extensions==1.1.0
 + nest-asyncio==1.6.0
 + networkx==3.5
 + nodeenv==1.9.1
 + numba==0.61.2
 + numcodecs==0.16.1
 + numpy==1.26.4
 + numpydantic==1.6.9
 + nvidia-cublas-cu12==12.6.4.1
 + nvidia-cuda-cupti-cu12==12.6.80
 + nvidia-cuda-nvcc-cu12==12.9.41
 + nvidia-cuda-nvrtc-cu12==12.6.77
 + nvidia-cuda-runtime-cu12==12.6.77
 + nvidia-cudnn-cu12==9.5.1.17
 + nvidia-cufft-cu12==11.3.0.4
 + nvidia-cufile-cu12==1.11.1.6
 + nvidia-curand-cu12==10.3.7.77
 + nvidia-cusolver-cu12==11.7.1.2
 + nvidia-cusparse-cu12==12.5.4.2
 + nvidia-cusparselt-cu12==0.6.3
 + nvidia-ml-py==12.575.51
 + nvidia-nccl-cu12==2.26.2
 + nvidia-nvjitlink-cu12==12.6.85
 + nvidia-nvtx-cu12==12.6.77
 + oauthlib==3.2.2
 + omegaconf==2.3.0
 + opencv-python==4.11.0.86
 + opencv-python-headless==4.11.0.86
 + openpi==0.1.0 (from file:///workspace/openpi)
 + openpi-client==0.1.0 (from file:///workspace/openpi/packages/openpi-client)
 + opt-einsum==3.4.0
 + optax==0.2.4
 + orbax-checkpoint==0.11.13
 + orderly-set==5.4.1
 + packaging==25.0
 + pandas==2.2.3
 + parso==0.8.4
 + pexpect==4.9.0
 + pfzy==0.3.4
 + pillow==11.2.1
 + platformdirs==4.3.8
 + pluggy==1.6.0
 + polars==1.30.0
 + pre-commit==4.2.0
 + prompt-toolkit==3.0.51
 + propcache==0.3.1
 + proto-plus==1.26.1
 + protobuf==4.25.8
 + psutil==7.0.0
 + ptyprocess==0.7.0
 + pure-eval==0.2.3
 + pyarrow==20.0.0
 + pyasn1==0.6.1
 + pyasn1-modules==0.4.2
 + pycparser==2.22
 + pydantic==2.11.5
 + pydantic-core==2.33.2
 + pygments==2.19.1
 + pymunk==7.0.0
 + pynput==1.8.1
 + pynvml==12.0.0
 + pyopengl==3.1.9
 + pyparsing==3.2.3
 + pysocks==1.7.1
 + pytest==8.3.5
 + python-dateutil==2.9.0.post0
 + python-xlib==0.33
 + pytz==2025.2
 + pyyaml==6.0.2
 + pyyaml-include==1.4.1
 + pyzmq==26.4.0
 + regex==2024.11.6
 + requests==2.32.3
 + requests-oauthlib==2.0.0
 + rerun-sdk==0.23.1
 + rich==14.0.0
 + rsa==4.9.1
 + ruff==0.11.12
 + safetensors==0.5.3
 + scipy==1.15.3
 + sentencepiece==0.2.0
 + sentry-sdk==2.29.1
 + setproctitle==1.3.6
 + setuptools==80.9.0
 + shtab==1.7.2
 + simplejson==3.20.1
 + six==1.17.0
 + smmap==5.0.2
 + soupsieve==2.7
 + stack-data==0.6.3
 + sympy==1.14.0
 + tensorstore==0.1.74
 + termcolor==3.1.0
 + tokenizers==0.21.1
 + toml==0.10.2
 + toolz==1.0.0
 + torch==2.7.1
 + torchcodec==0.4.0
 + torchvision==0.22.1
 + tornado==6.5.1
 + tqdm==4.67.1
 + tqdm-loggable==0.2
 + traitlets==5.14.3
 + transformers==4.53.2
 + treescope==0.1.9
 + triton==3.3.1
 + typeguard==4.4.2
 + typing-extensions==4.13.2
 + typing-inspect==0.9.0
 + typing-inspection==0.4.1
 + tyro==0.9.22
 + tzdata==2025.2
 + urllib3==2.4.0
 + virtualenv==20.31.2
 + wadler-lindig==0.1.6
 + wandb==0.19.11
 + wcwidth==0.2.13
 + websockets==15.0.1
 + werkzeug==3.1.3
 + widgetsnbextension==4.0.14
 + wrapt==1.14.1
 + xxhash==3.5.0
 + yarl==1.20.0
 + zarr==3.0.8
 + zipp==3.22.0
    (21 秒)

=== 3f. addon・設定・assets (pi05_mecharm_deep_even10_lora)
installed: policies/mecharm_policy.py + config.py (pi05_mecharm_lora)
installed: pi05_mecharm_trim_lora (repo_id=local/mecharm_redblock_box_teach_v1_trim2, 4,000 steps)
installed: pi05_mecharm_deep_even10_lora (repo_id=local/mecharm_redblock_box_deep_even10_v1, 3000 steps, decay_steps 5000, init=π0.5 base)
    (3 秒)

=== 4. サーバーの起動
[SERVE ] --model deep1: --policy.config=pi05_mecharm_deep_even10_lora --policy.dir=/workspace/ckpt/deep_v1_2999
INFO:websockets.server:server listening on 0.0.0.0:8000
    (53 秒)

=== 5. Mac の 8000 番へのトンネル
[TUNNEL] localhost:8000 → Pod の 8000 番 (pid 61370. このスクリプトが終わると閉じる)
    (3 秒)

=== 6. デーモンの状態
--- process
506 /usr/bin/python3 -u /home/ubuntu/mecharm_daemon.py
--- 5555
LISTEN   0         1                   0.0.0.0:5555             0.0.0.0:*       
--- systemd
enabled: enabled
active:  active
    (2 秒)

=== 7. HOME → 爪を開く → 位置合わせ (1 回目)
[INIT] daemon 接続中...
  [Pi] fresh_mode=1
[INIT] fresh_mode=1
[PRE]  現在姿勢: [-0.26, 92.46, 1.84, -2.19, -96.76, -29.53]
[SEND] HOME へ移動指令
[WAIT] t=0.6s  max_err=82.35°
[WAIT] t=1.1s  max_err=66.35°
[WAIT] t=1.6s  max_err=51.06°
[WAIT] t=2.1s  max_err=36.29°
[WAIT] t=2.7s  max_err=20.91°
[WAIT] t=3.2s  max_err=5.62°
[WAIT] t=3.7s  max_err=1.23°
[DONE] HOME 到達 max_err=1.23°
[POST] 最終姿勢: [-0.17, 1.23, 1.05, -0.52, 1.14, -29.53]
  [Pi] fresh_mode=0
[EXIT] fresh_mode=0, socket close
  [Pi] GRIPPER: OPEN result= None
[REF   ] 基準画像: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_194818/frames/000000.jpg
[REF   ] お手本 run_0001_20261019_194818 (つかんだ瞬間の j0 = +13.2°)
[REF   ] 重なった画像のブロックに実物のブロックを合わせると, お手本と同じ位置に置ける
[TILT  ] 傾きのお手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  中央値 -0.61° (測れた 50/50 本. プラス=右下がり)
[TILT  ] 傾きはソフトでそろえる: ライブをお手本の中央値との差だけ回して重ねる (AI に渡す画像と同じ. 回す角度が 3° を超えたら画面の色を変える. 止めない)
[ALIGN ] |dy| <= 20 なら OK. アームは HOME に置いた状態で合わせる. q または Esc で終了 (ファイルには何も書かない)
[ALIGN ] dy=  -0.1  dx=  +0.8  response=0.826  → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  傾き +0.62° → -0.61° にソフトでそろえる (-1.23° 回して重ねる. プラス=右下がり)
[ALIGN ] dy=  +6.0  dx=  +4.2  response=0.071  → 測れない (相関が弱く dx/dy が当てにならない (response=0.071 < 0.20))  傾き -8.30° → -0.61° にソフトでそろえる (+7.69° 回して重ねる. プラス=右下がり) → 回す角度が 3° を超える (TILT DIFFERS. iPhone の付け方を見直す)
[ALIGN ] dy=  +4.5  dx=  -3.5  response=0.009  → 測れない (相関が弱く dx/dy が当てにならない (response=0.009 < 0.20))  傾き -8.30° → -0.61° にソフトでそろえる (+7.69° 回して重ねる. プラス=右下がり) → 回す角度が 3° を超える (TILT DIFFERS. iPhone の付け方を見直す)
[ALIGN ] dy=  +5.5  dx=  +5.6  response=-0.022  → 測れない (相関が弱く dx/dy が当てにならない (response=-0.022 < 0.20))  傾き -7.99° → -0.61° にソフトでそろえる (+7.37° 回して重ねる. プラス=右下がり) → 回す角度が 3° を超える (TILT DIFFERS. iPhone の付け方を見直す)
[ALIGN ] dy=  +0.4  dx=  +1.5  response=0.875  → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  傾き +0.53° → -0.61° にソフトでそろえる (-1.15° 回して重ねる. プラス=右下がり)
[ALIGN ] dy=  +7.1  dx=  +3.5  response=0.074  → 測れない (相関が弱く dx/dy が当てにならない (response=0.074 < 0.20))  傾き -9.45° → -0.61° にソフトでそろえる (+8.83° 回して重ねる. プラス=右下がり) → 回す角度が 3° を超える (TILT DIFFERS. iPhone の付け方を見直す)
[ALIGN ] dy=  +0.2  dx=  +1.1  response=0.752  → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  傾き +0.50° → -0.61° にソフトでそろえる (-1.12° 回して重ねる. プラス=右下がり)
[DONE  ] 終了 (何も書いていません)
    (170 秒)

=== 8. 実機推論 (1 回目)
[WAIT  ] 送ったあと 0.0s / 爪のあと 0.8s (開くときの daemon 0.0s) / 止まるのを待つ目安 0.8s
[GRIP  ] 閉じる値 25  締め直し r は 22 と 25 を交互に送る
[SAFE  ] 安全範囲 (学習 50 本 ±10°): j0[-72,+55]  j1[-9,+88]  j2[-45,+52]  j3[-18,+27]  j4[-87,+10]  j5[-51,-15]
[SAFE  ] 肩 j1 の上限は J1_SAFE_MAX=88° で切った (これを超える指令はアームが黙って捨てる / LOG352)  元データ: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes
[MODE  ] EXECUTE  steps/chunk=10  max_move=20.0°  log=/Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_160851
[MODEL ] deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)
[CAMERA] 写り方 dy=-6.1 dx=+7.8 response=0.555 → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  回したあとの画像で測った
[CAMERA] 基準: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_183454/frames/000000.jpg
[CAMERA] 傾き +0.57° → -0.61° にそろえて渡す (-1.19° 回す. プラス=右下がり) (お手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  測れた 50/50 本・今 5/5 コマ)
[POLICY] 接続 ws://localhost:8000  metadata={}
[RECORD] 上カメラと実測角を 1 秒 10 コマで記録します → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_160851/recording
[GRIP  ] 開始時のグリッパー開放は, 最初の目標が検査を通ってから行います (爪が閉じていたら手で開けておく)
[STATE ] 実測を渡す (直前に送った目標がない)
[000] 推論 63593ms  10ステップ先  Δ= +2.6 +15.3  +8.3  -1.2 -20.5  +0.9
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[GRIP  ] 開始時にグリッパーを開く
  [Pi] GRIPPER: OPEN result= None
[SEND] pi05_chunk000_via02: send_angles (single) [  +0.34,   +5.17,   +2.83,   -0.24,   -4.75,  -29.76]  speed=20 wait=0.0s  (pre max|Δ|=4.49°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.68° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.5  +3.9  +1.8  +0.0  -4.5  -0.1  着いた
[SEND] pi05_chunk000_via05: send_angles (single) [  +1.12,   +9.77,   +5.21,   -0.33,  -11.03,  -29.61]  speed=20 wait=0.0s  (pre max|Δ|=7.96°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.37° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.3  +3.7  +1.4  -0.1  -8.0  +0.1  着いた
[SEND] pi05_chunk000_via08: send_angles (single) [  +2.21,  +15.21,   +8.52,   -1.20,  -18.81,  -28.98]  speed=20 wait=0.0s  (pre max|Δ|=9.06°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.41° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.5  +4.4  +2.4  -1.0  -9.1  +0.7  着いた
[SEND] pi05_chunk000_via09: send_angles (single) [  +2.40,  +16.53,   +9.38,   -1.48,  -20.80,  -28.77]  speed=20 wait=0.0s  (pre max|Δ|=3.40°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.13° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.7  +0.4  -0.0  -1.2  -3.4  +0.9  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.05°（関節 j3）
[001] 推論 806ms  10ステップ先  Δ= +4.7 +13.5  +6.8  -1.8 -19.8  +2.3
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk001_via02: send_angles (single) [  +3.63,  +20.08,  +11.46,   -1.88,  -25.67,  -28.12]  speed=20 wait=0.0s  (pre max|Δ|=5.90°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.92° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.3  +2.7  +1.4  -1.3  -5.9  +1.4  着いた
[SEND] pi05_chunk001_via05: send_angles (single) [  +5.16,  +24.34,  +13.81,   -2.13,  -31.47,  -27.67]  speed=20 wait=0.0s  (pre max|Δ|=6.60°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.97° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.8  +3.3  +1.6  -0.5  -6.6  +1.2  着いた
[SEND] pi05_chunk001_via08: send_angles (single) [  +6.76,  +29.35,  +16.44,   -2.27,  -37.89,  -27.49]  speed=20 wait=0.0s  (pre max|Δ|=6.96°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.97° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.8  +4.0  +1.9  -0.4  -7.0  +0.8  着いた
[SEND] pi05_chunk001_via09: send_angles (single) [  +7.04,  +30.91,  +16.90,   -2.22,  -39.60,  -27.44]  speed=20 wait=0.0s  (pre max|Δ|=2.52°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.81° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.4  +0.6  -0.2  -0.2  -2.5  +0.6  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.99°（関節 j1）
[002] 推論 803ms  10ステップ先  Δ= +2.9  +7.5  +3.5  +0.2 -10.5  +0.6
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk002_via02: send_angles (single) [  +7.83,  +33.59,  +18.32,   -2.02,  -42.90,  -27.40]  speed=20 wait=0.0s  (pre max|Δ|=3.53°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.86° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.2  +1.7  +0.7  -0.0  -3.5  +0.3  着いた
[SEND] pi05_chunk002_via05: send_angles (single) [  +8.75,  +36.53,  +19.48,   -1.92,  -46.22,  -27.38]  speed=20 wait=0.0s  (pre max|Δ|=3.60°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.91° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.0  +2.1  +0.6  +0.0  -3.6  +0.2  着いた
[SEND] pi05_chunk002_via08: send_angles (single) [  +9.38,  +39.10,  +20.92,   -1.85,  -49.40,  -27.36]  speed=20 wait=0.0s  (pre max|Δ|=3.53°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.89° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.6  +1.7  +1.0  +0.2  -3.5  +0.2  着いた
[SEND] pi05_chunk002_via09: send_angles (single) [  +9.45,  +39.37,  +21.09,   -1.83,  -49.91,  -27.34]  speed=20 wait=0.0s  (pre max|Δ|=1.05°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.79° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.1  -0.6  -0.3  +0.2  -1.1  +0.2  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.79°（関節 j1）
[003] 推論 2216ms  10ステップ先  Δ= +2.8  +3.1  +2.6  +0.3  -6.2  +0.3
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk003_via02: send_angles (single) [ +10.83,  +40.36,  +22.18,   -1.78,  -52.12,  -27.34]  speed=20 wait=0.0s  (pre max|Δ|=2.82°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.94° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.5  +0.2  +0.7  +0.1  -2.8  +0.2  着いた
[SEND] pi05_chunk003_via05: send_angles (single) [ +11.63,  +41.01,  +22.96,   -1.64,  -53.14,  -27.31]  speed=20 wait=0.0s  (pre max|Δ|=1.00°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.91° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.0  -0.3  +0.4  +0.2  -0.6  +0.3  着いた
[SEND] pi05_chunk003_via08: send_angles (single) [ +12.03,  +42.53,  +23.79,   -1.57,  -54.70,  -27.30]  speed=20 wait=0.0s  (pre max|Δ|=1.71°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.97° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.3  +0.6  +0.4  +0.3  -1.7  +0.2  着いた
       [VIA   ] step  9  Δ= +0.3  -0.2  +0.2  +0.3  -0.4  +0.2  送らず (0.5° 以下)
[STATE ] 目標を渡す  目標と実測の最大差 0.41°（関節 j4）
[004] 推論 1222ms  10ステップ先  Δ= +3.8  +4.6  +1.8  +0.1  -5.9  +0.3
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk004_via02: send_angles (single) [ +13.31,  +45.27,  +25.39,   -1.58,  -57.69,  -27.21]  speed=20 wait=0.0s  (pre max|Δ|=2.59°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.96° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.5  +1.8  +1.5  +0.3  -2.6  +0.3  着いた
[SEND] pi05_chunk004_via05: send_angles (single) [ +14.62,  +46.68,  +26.14,   -1.58,  -59.91,  -27.21]  speed=20 wait=0.0s  (pre max|Δ|=2.17°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.04° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.6  +0.4  +0.4  -0.1  -2.2  +0.2  着いた
[SEND] pi05_chunk004_via08: send_angles (single) [ +15.01,  +47.21,  +25.79,   -1.77,  -60.65,  -27.22]  speed=20 wait=0.0s  (pre max|Δ|=1.15°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.04° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.5  -0.5  +0.0  -0.2  -1.1  +0.2  着いた
[SEND] pi05_chunk004_via09: send_angles (single) [ +15.61,  +48.06,  +25.73,   -1.76,  -60.95,  -27.23]  speed=20 wait=0.0s  (pre max|Δ|=1.11°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.89° (tol 3.0°)
       [VIA   ] step  9  Δ= +1.1  -0.2  -0.0  -0.3  -0.8  +0.2  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.89°（関節 j1）
[005] 推論 1042ms  10ステップ先  Δ= +0.2 +13.7  -9.9  -0.6  -1.8  -0.5
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk005_via02: send_angles (single) [ +15.80,  +53.48,  +22.28,   -1.89,  -61.48,  -27.30]  speed=20 wait=0.0s  (pre max|Δ|=4.53°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.92° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.3  +4.5  -3.5  -0.4  -0.7  +0.1  着いた
[SEND] pi05_chunk005_via05: send_angles (single) [ +15.85,  +59.56,  +18.40,   -1.99,  -62.38,  -27.58]  speed=20 wait=0.0s  (pre max|Δ|=5.16°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.78° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.3  +5.2  -4.8  -0.5  -1.1  -0.2  着いた
[SEND] pi05_chunk005_via08: send_angles (single) [ +15.86,  +62.53,  +16.06,   -2.03,  -62.46,  -27.85]  speed=20 wait=0.0s  (pre max|Δ|=3.27°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.52° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.3  +2.1  -3.3  -0.5  +0.6  -0.4  着いた
[SEND] pi05_chunk005_via09: send_angles (single) [ +15.79,  +62.68,  +15.87,   -2.04,  -62.57,  -27.93]  speed=20 wait=0.0s  (pre max|Δ|=1.59°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.59° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.2  -0.6  -0.8  -0.6  +1.6  -0.4  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.59°（関節 j4）
[006] 推論 809ms  3ステップ先  Δ= +0.1  -0.3  -0.8  -0.5  +1.3  -0.4  グリッパー→閉
       [VIA   ] 中継点 1 点: step 2
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk006_via02: send_angles (single) [ +15.63,  +63.06,  +15.86,   -1.93,  -62.86,  -28.03]  speed=20 wait=0.0s  (pre max|Δ|=1.30°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.38° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.1  -0.3  -0.8  -0.4  +1.3  -0.4  着いた
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[STATE ] 目標を渡す  目標と実測の最大差 2.44°（関節 j4）
[007] 推論 2232ms  10ステップ先  Δ= +1.2  -6.3  +4.9  -0.8  +2.2  +0.9
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk007_via02: send_angles (single) [ +15.90,  +61.03,  +18.01,   -2.18,  -62.67,  -27.63]  speed=20 wait=0.0s  (pre max|Δ|=2.51°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.11° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.7  -2.5  +2.2  -0.5  +0.5  +0.6  着いた
[SEND] pi05_chunk007_via05: send_angles (single) [ +16.06,  +58.95,  +20.73,   -2.21,  -62.98,  -27.29]  speed=20 wait=0.0s  (pre max|Δ|=4.30°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.78° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.9  -2.7  +4.3  -0.4  -2.8  +0.7  着いた
[SEND] pi05_chunk007_via08: send_angles (single) [ +16.25,  +57.83,  +21.55,   -2.18,  -63.27,  -27.28]  speed=20 wait=0.0s  (pre max|Δ|=3.07°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.89° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.7  -1.6  +2.1  -0.2  -3.1  +0.3  着いた
[SEND] pi05_chunk007_via09: send_angles (single) [ +16.48,  +57.37,  +21.35,   -2.07,  -63.15,  -27.39]  speed=20 wait=0.0s  (pre max|Δ|=2.69°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.60° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.8  -1.2  +1.0  -0.0  -2.8  +0.1  着いた
[STATE ] 目標を渡す  目標と実測の最大差 2.60°（関節 j4）
[008] 推論 808ms  10ステップ先  Δ= +3.4  -4.6  +1.0  +0.8  -3.4  -0.2
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk008_via02: send_angles (single) [ +16.55,  +53.05,  +25.67,   -2.46,  -62.84,  -26.64]  speed=20 wait=0.0s  (pre max|Δ|=5.39°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.11° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.9  -5.4  +5.2  -0.5  -1.7  +0.1  着いた
[SEND] pi05_chunk008_via05: send_angles (single) [ +16.81,  +50.50,  +27.96,   -2.44,  -63.77,  -26.12]  speed=20 wait=0.0s  (pre max|Δ|=4.08°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.87° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.6  -4.1  +2.8  -0.2  -3.1  -0.3  着いた
[SEND] pi05_chunk008_via08: send_angles (single) [ +17.54,  +53.17,  +23.86,   -1.91,  -64.33,  -27.09]  speed=20 wait=0.0s  (pre max|Δ|=4.00°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.37° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.3  +0.9  -4.0  +0.8  -3.4  -1.8  着いた
[SEND] pi05_chunk008_via09: send_angles (single) [ +19.02,  +53.82,  +21.40,   -1.20,  -63.97,  -27.71]  speed=20 wait=0.0s  (pre max|Δ|=3.38°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.76° (tol 3.0°)
       [VIA   ] step  9  Δ= +1.5  -0.2  -3.4  +1.2  -2.0  -2.1  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.84°（関節 j4）
[009] 推論 1102ms  10ステップ先  Δ= +6.2  +0.2  -9.9  +3.9  +0.6  -4.0
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk009_via02: send_angles (single) [ +21.61,  +54.04,  +18.42,   -0.22,  -62.95,  -28.51]  speed=20 wait=0.0s  (pre max|Δ|=3.90°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.35° (tol 3.0°)
       [VIA   ] step  2  Δ= +2.7  -0.7  -3.9  +1.5  -0.4  -1.7  着いた
[SEND] pi05_chunk009_via05: send_angles (single) [ +24.26,  +54.48,  +15.25,   +0.72,  -61.87,  -29.34]  speed=20 wait=0.0s  (pre max|Δ|=3.99°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.06° (tol 3.0°)
       [VIA   ] step  5  Δ= +2.7  -0.4  -4.0  +1.2  +0.8  -1.6  着いた
[SEND] pi05_chunk009_via08: send_angles (single) [ +25.28,  +54.85,  +12.95,   +1.67,  -61.53,  -30.10]  speed=20 wait=0.0s  (pre max|Δ|=3.22°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.13° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.2  -0.7  -3.2  +1.2  +1.1  -1.4  着いた
[SEND] pi05_chunk009_via09: send_angles (single) [ +25.33,  +54.98,  +12.47,   +1.84,  -61.52,  -30.25]  speed=20 wait=0.0s  (pre max|Δ|=1.50°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.41° (tol 3.0°)
       [VIA   ] step  9  Δ= -0.1  -0.9  -1.5  +0.7  +1.1  -0.6  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.41°（関節 j2）
[010] 推論 2418ms  10ステップ先  Δ= +0.8  -2.0  -7.9  +0.8  +1.7  -0.6
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > p
[PAD   ] 補正 1 (推論 10 のあと). ゲームパッドで直して, □ で推論に戻ります
コントローラーを待っています（5 秒でつながらなければ推論の入力待ちに戻ります）。□ で推論に戻ります。Ctrl-C で補正を抜けます
/opt/homebrew/Caskroom/miniforge/base/lib/python3.12/site-packages/pygame/pkgdata.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
  from pkg_resources import resource_stream, resource_exists
[接続] PS4 Controller  GUID=03008fe54c050000cc09000000016800  軸=6  ボタン=16
[長押し] j1 を +55.98° → +48.77° まで（目標を 1 回送信。止めます）
[長押し] j1 を +44.73° → +39.81° まで（目標を 1 回送信。止めます）
[PAD   ] 補正 1 を終えました (14.0 秒, 送った目標 2 回). この姿勢から推論に戻ります
[STATE ] 実測を渡す (直前に送った目標がない)
[011] 推論 1169ms  10ステップ先  Δ= +9.0 +10.2  -4.3  +1.4  -2.0  -1.0
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk011_via02: send_angles (single) [ +28.91,  +33.33,  +11.81,   +1.99,  -60.36,  -30.17]  speed=20 wait=0.0s  (pre max|Δ|=3.60°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.07° (tol 3.0°)
       [VIA   ] step  2  Δ= +3.6  +1.9  -2.1  +0.5  +0.5  -0.4  着いた
[SEND] pi05_chunk011_via05: send_angles (single) [ +32.09,  +37.19,  +10.42,   +2.70,  -61.54,  -30.67]  speed=20 wait=0.0s  (pre max|Δ|=3.29°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.44° (tol 3.0°)
       [VIA   ] step  5  Δ= +3.3  +2.7  -3.3  +0.6  -0.3  -0.9  着いた
[SEND] pi05_chunk011_via08: send_angles (single) [ +34.02,  +40.80,   +9.62,   +2.82,  -62.70,  -30.71]  speed=20 wait=0.0s  (pre max|Δ|=2.48°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.24° (tol 3.0°)
       [VIA   ] step  8  Δ= +2.0  +2.5  -2.2  -0.1  -1.2  -0.5  着いた
[SEND] pi05_chunk011_via09: send_angles (single) [ +34.34,  +41.68,   +9.62,   +2.86,  -62.91,  -30.75]  speed=20 wait=0.0s  (pre max|Δ|=1.63°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.63° (tol 3.0°)
       [VIA   ] step  9  Δ= -0.0  -0.2  -1.7  +0.1  -0.5  -0.4  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.63°（関節 j2）
[012] 推論 2365ms  10ステップ先  Δ= +4.9 +10.9  -4.8  +0.0  -5.9  -0.7
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk012_via02: send_angles (single) [ +38.71,  +44.20,   +7.44,   +2.92,  -62.30,  -30.93]  speed=20 wait=0.0s  (pre max|Δ|=4.44°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.87° (tol 3.0°)
       [VIA   ] step  2  Δ= +4.4  +1.4  -3.8  -0.2  +0.5  -0.5  着いた
[SEND] pi05_chunk012_via05: send_angles (single) [ +39.22,  +48.80,   +6.81,   +2.99,  -65.22,  -30.97]  speed=20 wait=0.0s  (pre max|Δ|=3.72°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.71° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.5  +3.7  -2.1  -0.3  -2.4  -0.5  着いた
[SEND] pi05_chunk012_via08: send_angles (single) [ +39.02,  +53.13,   +6.45,   +3.19,  -68.39,  -31.01]  speed=20 wait=0.0s  (pre max|Δ|=3.97°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.81° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.4  +3.3  -1.9  -0.1  -4.0  -0.3  着いた
[SEND] pi05_chunk012_via09: send_angles (single) [ +39.14,  +53.65,   +6.49,   +3.19,  -68.78,  -31.01]  speed=20 wait=0.0s  (pre max|Δ|=1.33°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.33° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.5  -0.7  -1.3  -0.1  -0.9  -0.3  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.33°（関節 j2）
[013] 推論 1003ms  4ステップ先  Δ= +0.6  +3.0  -2.0  +0.2  -3.0  -0.4  グリッパー→開
       [VIA   ] 中継点 2 点: step 2, 3
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk013_via02: send_angles (single) [ +39.35,  +57.76,   +5.86,   +3.50,  -71.14,  -31.09]  speed=20 wait=0.0s  (pre max|Δ|=3.03°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.96° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.6  +3.0  -2.0  +0.2  -3.0  -0.4  着いた
[SEND] pi05_chunk013_via03: send_angles (single) [ +39.35,  +57.76,   +5.86,   +3.50,  -71.14,  -31.09]  speed=20 wait=0.0s  (pre max|Δ|=1.25°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.25° (tol 3.0°)
       [VIA   ] step  3  Δ= +0.6  -1.1  -1.3  +0.2  -0.2  -0.3  着いた
  [Pi] GRIPPER: OPEN result= None
[STATE ] 目標を渡す  目標と実測の最大差 1.25°（関節 j2）
[014] 推論 811ms  10ステップ先  Δ= -2.0 -16.9  -1.5  -0.3 +17.1  -0.6
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk014_via02: send_angles (single) [ +39.46,  +53.58,   +5.79,   +3.38,  -65.77,  -31.20]  speed=20 wait=0.0s  (pre max|Δ|=5.30°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.78° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.7  -5.3  -1.3  +0.1  +5.1  -0.4  着いた
[SEND] pi05_chunk014_via05: send_angles (single) [ +39.18,  +48.78,   +5.80,   +3.26,  -60.50,  -31.29]  speed=20 wait=0.0s  (pre max|Δ|=7.96°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.86° (tol 3.0°)
       [VIA   ] step  5  Δ= -0.1  -6.7  -1.3  -0.1  +8.0  -0.5  着いた
[SEND] pi05_chunk014_via08: send_angles (single) [ +37.65,  +43.68,   +5.64,   +3.13,  -55.40,  -31.35]  speed=20 wait=0.0s  (pre max|Δ|=7.44°, post-send timeout 8.0s)
  [WARN] daemon_get_angles failed: daemon error: get_angles returned empty/invalid after 12 retries
[SEND] arrival OK: max|Δ|=1.84° (tol 3.0°)
       [VIA   ] step  8  Δ= -1.6  -7.1  -1.5  -0.3  +7.5  -0.3  着いた
[SEND] pi05_chunk014_via09: send_angles (single) [ +36.79,  +42.02,   +5.58,   +3.04,  -53.83,  -31.36]  speed=20 wait=0.0s  (pre max|Δ|=3.50°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.92° (tol 3.0°)
       [VIA   ] step  9  Δ= -0.8  -3.5  -1.5  -0.5  +2.8  -0.3  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.92°（関節 j1）
[015] 推論 933ms  10ステップ先  Δ=-15.3 -21.8  -1.9  -2.7 +25.1  -0.2
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > h
[SEND] pi05_home: send_angles (single) [  +0.00,   +0.00,   +0.00,   +0.00,   +0.00,  -30.00]  speed=20 wait=0.0s  (pre max|Δ|=54.58°, post-send timeout 10.2s)
[SEND] arrival OK: max|Δ|=1.75° (tol 2.5°)
[RECORD] 記録終了 コマ 2562 (8.88 fps) 実測角 2683 回 (9.3 Hz)  電流 2700 行 (read_servo)  → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_160851/recording
       結果: y=成功 / n=失敗 / Enter=記録しない > y
[GRIP  ] 閉じて 1 秒後の爪の電流 (中央値): 1 回目 751 (10 行)  (閉じる値 25)
[DONE  ] 記録: /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_160851
[CLIENT] run_policy_client.py が終わった (終了コード 0. モデル deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1))
    (300 秒)
もう1回やりますか? y=やる / 他=終わる > y

=== 7. HOME → 爪を開く → 位置合わせ (2 回目)
[INIT] daemon 接続中...
  [Pi] fresh_mode=1
[INIT] fresh_mode=1
[PRE]  現在姿勢: [-0.17, 1.49, 1.75, 0.26, -0.26, -30.32]
[SEND] HOME へ移動指令
[WAIT] t=0.6s  max_err=1.75°
[DONE] HOME 到達 max_err=1.75°
[POST] 最終姿勢: [-0.17, 1.49, 1.75, 0.26, -0.26, -30.23]
  [Pi] fresh_mode=0
[EXIT] fresh_mode=0, socket close
  [Pi] GRIPPER: OPEN result= None
[REF   ] 基準画像: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_194818/frames/000000.jpg
[REF   ] お手本 run_0001_20261019_194818 (つかんだ瞬間の j0 = +13.2°)
[REF   ] 重なった画像のブロックに実物のブロックを合わせると, お手本と同じ位置に置ける
[TILT  ] 傾きのお手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  中央値 -0.61° (測れた 50/50 本. プラス=右下がり)
[TILT  ] 傾きはソフトでそろえる: ライブをお手本の中央値との差だけ回して重ねる (AI に渡す画像と同じ. 回す角度が 3° を超えたら画面の色を変える. 止めない)
[ALIGN ] |dy| <= 20 なら OK. アームは HOME に置いた状態で合わせる. q または Esc で終了 (ファイルには何も書かない)
[ALIGN ] dy=  +0.4  dx=  +0.9  response=0.942  → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  傾き +0.53° → -0.61° にソフトでそろえる (-1.15° 回して重ねる. プラス=右下がり)
[DONE  ] 終了 (何も書いていません)
    (10 秒)

=== 8. 実機推論 (2 回目)
[WAIT  ] 送ったあと 0.0s / 爪のあと 0.8s (開くときの daemon 0.0s) / 止まるのを待つ目安 0.8s
[GRIP  ] 閉じる値 25  締め直し r は 22 と 25 を交互に送る
[SAFE  ] 安全範囲 (学習 50 本 ±10°): j0[-72,+55]  j1[-9,+88]  j2[-45,+52]  j3[-18,+27]  j4[-87,+10]  j5[-51,-15]
[SAFE  ] 肩 j1 の上限は J1_SAFE_MAX=88° で切った (これを超える指令はアームが黙って捨てる / LOG352)  元データ: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes
[MODE  ] EXECUTE  steps/chunk=10  max_move=20.0°  log=/Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161402
[MODEL ] deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)
[CAMERA] 写り方 dy=-5.8 dx=+7.8 response=0.555 → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  回したあとの画像で測った
[CAMERA] 基準: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_183454/frames/000000.jpg
[CAMERA] 傾き +0.53° → -0.61° にそろえて渡す (-1.15° 回す. プラス=右下がり) (お手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  測れた 50/50 本・今 5/5 コマ)
[POLICY] 接続 ws://localhost:8000  metadata={}
[RECORD] 上カメラと実測角を 1 秒 10 コマで記録します → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161402/recording
[GRIP  ] 開始時のグリッパー開放は, 最初の目標が検査を通ってから行います (爪が閉じていたら手で開けておく)
[STATE ] 実測を渡す (直前に送った目標がない)
[000] 推論 2192ms  10ステップ先  Δ= +5.1 +17.4 +10.9  -1.6 -23.4  +2.3
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[GRIP  ] 開始時にグリッパーを開く
  [Pi] GRIPPER: OPEN result= None
[SEND] pi05_chunk000_via02: send_angles (single) [  +0.95,   +5.58,   +4.51,   +0.05,   -5.39,  -29.76]  speed=20 wait=0.0s  (pre max|Δ|=5.13°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.06° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.1  +4.1  +2.8  -0.2  -5.1  +0.5  着いた
[SEND] pi05_chunk000_via05: send_angles (single) [  +2.52,  +10.90,   +7.85,   -0.46,  -12.35,  -29.00]  speed=20 wait=0.0s  (pre max|Δ|=9.02°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.54° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.8  +4.4  +2.3  -0.6  -9.0  +1.0  着いた
[SEND] pi05_chunk000_via08: send_angles (single) [  +4.44,  +16.96,  +11.49,   -1.13,  -21.02,  -28.15]  speed=20 wait=0.0s  (pre max|Δ|=10.21°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.34° (tol 3.0°)
       [VIA   ] step  8  Δ= +2.3  +5.1  +2.7  -0.8 -10.2  +1.0  着いた
[SEND] pi05_chunk000_via09: send_angles (single) [  +4.90,  +18.88,  +12.61,   -1.32,  -23.69,  -27.90]  speed=20 wait=0.0s  (pre max|Δ|=4.01°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.02° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.7  +1.0  +0.4  +0.3  -4.0  +0.5  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.02°（関節 j4）
[001] 推論 809ms  10ステップ先  Δ= +4.4 +14.5  +8.4  -1.2 -25.1  +1.3
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk001_via02: send_angles (single) [  +5.98,  +22.78,  +15.25,   -2.05,  -30.35,  -27.26]  speed=20 wait=0.0s  (pre max|Δ|=7.68°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.88° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.1  +2.9  +1.9  -0.4  -7.7  +0.9  着いた
[SEND] pi05_chunk001_via05: send_angles (single) [  +7.70,  +28.23,  +18.19,   -2.61,  -38.47,  -26.93]  speed=20 wait=0.0s  (pre max|Δ|=8.86°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.94° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.9  +4.5  +2.2  -0.5  -8.9  +0.5  着いた
[SEND] pi05_chunk001_via08: send_angles (single) [  +9.11,  +33.22,  +21.28,   -2.87,  -46.21,  -26.89]  speed=20 wait=0.0s  (pre max|Δ|=8.51°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.88° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.9  +4.1  +2.5  -0.3  -8.5  +0.3  着いた
[SEND] pi05_chunk001_via09: send_angles (single) [  +9.35,  +34.39,  +21.79,   -2.90,  -47.72,  -26.91]  speed=20 wait=0.0s  (pre max|Δ|=2.02°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.94° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.6  +0.3  -0.1  -0.1  -2.0  +0.2  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.94°（関節 j1）
[002] 推論 805ms  10ステップ先  Δ= +2.8 +10.2  +5.6  +0.5 -13.6  +0.2
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk002_via02: send_angles (single) [ +10.17,  +37.26,  +23.50,   -2.65,  -51.93,  -26.88]  speed=20 wait=0.0s  (pre max|Δ|=4.65°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.61° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.4  +1.9  +1.6  +0.1  -4.6  +0.2  着いた
[SEND] pi05_chunk002_via05: send_angles (single) [ +11.01,  +40.97,  +25.50,   -2.45,  -56.46,  -26.89]  speed=20 wait=0.0s  (pre max|Δ|=5.05°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.18° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.9  +2.7  +1.4  +0.2  -5.0  +0.2  着いた
[SEND] pi05_chunk002_via08: send_angles (single) [ +11.42,  +44.67,  +27.26,   -2.30,  -60.17,  -26.88]  speed=20 wait=0.0s  (pre max|Δ|=3.92°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.03° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.3  +2.7  +1.3  +0.3  -3.9  +0.2  着いた
[SEND] pi05_chunk002_via09: send_angles (single) [ +11.56,  +45.53,  +27.45,   -2.21,  -60.92,  -26.87]  speed=20 wait=0.0s  (pre max|Δ|=0.90°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.96° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.5  -0.2  -0.1  +0.2  -0.9  +0.2  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.96°（関節 j1）
[003] 推論 838ms  10ステップ先  Δ= +2.3 +15.1  -5.2  -0.1  -3.5  -0.1
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk003_via02: send_angles (single) [ +12.51,  +49.84,  +27.96,   -2.26,  -62.49,  -26.86]  speed=20 wait=0.0s  (pre max|Δ|=3.35°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.87° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.4  +3.3  +0.4  +0.2  -2.0  +0.2  着いた
[SEND] pi05_chunk003_via05: send_angles (single) [ +13.12,  +55.65,  +26.26,   -2.40,  -63.72,  -26.88]  speed=20 wait=0.0s  (pre max|Δ|=4.94°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.33° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.8  +4.9  -1.3  -0.0  -1.4  +0.2  着いた
[SEND] pi05_chunk003_via08: send_angles (single) [ +13.24,  +60.53,  +23.29,   -2.51,  -64.02,  -27.05]  speed=20 wait=0.0s  (pre max|Δ|=4.11°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.77° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.4  +4.1  -4.0  -0.6  +1.0  +0.1  着いた
[SEND] pi05_chunk003_via09: send_angles (single) [ +13.33,  +61.63,  +22.35,   -2.52,  -63.91,  -27.14]  speed=20 wait=0.0s  (pre max|Δ|=3.67°, post-send timeout 8.0s)

[SEND] arrival OK: max|Δ|=2.79° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.5  +0.5  -1.4  -0.5  +3.7  +0.3  着いた
[STATE ] 目標を渡す  目標と実測の最大差 2.79°（関節 j4）
[004] 推論 2212ms  1ステップ先  Δ= +0.0  +0.0  +0.0  +0.0  +0.0  +0.0  グリッパー→閉
       [VIA   ] 中継点 1 点: step 0
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
       [VIA   ] step  0  Δ= +0.0  +0.0  +0.2  +0.0  -0.4  +0.0  送らず (0.5° 以下)
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[STATE ] 実測を渡す (直前に送った目標がない)
[005] 推論 808ms  10ステップ先  Δ= +2.6  -6.4  +1.6  +0.4  +2.7  -0.2
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk005_via02: send_angles (single) [ +12.75,  +56.71,  +26.67,   -2.66,  -62.86,  -26.92]  speed=20 wait=0.0s  (pre max|Δ|=6.13°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.17° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.1  -5.4  +4.4  -0.5  +6.1  -0.1  着いた
[SEND] pi05_chunk005_via05: send_angles (single) [ +12.49,  +53.46,  +29.65,   -3.04,  -62.69,  -26.44]  speed=20 wait=0.0s  (pre max|Δ|=4.19°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.18° (tol 3.0°)
       [VIA   ] step  5  Δ= -0.2  -4.2  +4.0  +0.4  +2.3  +0.4  着いた
[SEND] pi05_chunk005_via08: send_angles (single) [ +13.80,  +54.61,  +26.34,   -2.17,  -63.73,  -27.18]  speed=20 wait=0.0s  (pre max|Δ|=2.13°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.24° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.1  +0.2  -2.1  +1.8  -1.0  -0.5  着いた
[SEND] pi05_chunk005_via09: send_angles (single) [ +15.39,  +55.33,  +23.83,   -1.56,  -63.10,  -27.83]  speed=20 wait=0.0s  (pre max|Δ|=2.80°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.90° (tol 3.0°)
       [VIA   ] step  9  Δ= +1.9  +0.4  -2.8  +1.4  -1.4  -1.0  着いた
[STATE ] 目標を渡す  目標と実測の最大差 2.90°（関節 j4）
[006] 推論 2431ms  10ステップ先  Δ= +9.6  -0.8  -7.6  +3.3  +0.3  -2.9
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk006_via02: send_angles (single) [ +16.14,  +54.63,  +23.97,   -1.06,  -63.33,  -28.15]  speed=20 wait=0.0s  (pre max|Δ|=3.13°, post-send timeout 8.0s)
[SEND] [TIMEOUT] 8.0s 内に未到達: max|Δ|=3.13° actual=[ +15.55,  +55.63,  +24.08,   -0.96,  -60.20,  -27.86]
[RETRY] 1回目送信が無視された（pre→last max|Δ| 0.88°）。5.0s 待って再送します
[RETRY] pi05_chunk006_via02 (retry): send_angles (single) [ +16.14,  +54.63,  +23.97,   -1.06,  -63.33,  -28.15]  speed=20 wait=0.0s  (pre max|Δ|=3.13°, post-send timeout 8.0s)
[RETRY] 失敗。ABORT へ. 8.0s 内に未到達: max|Δ|=3.13° actual=[ +15.55,  +55.54,  +24.08,   -0.96,  -60.20,  -27.86]
       [VIA   ] step  2  Δ= +1.2  -1.1  -0.1  +0.8  -3.1  -0.6  未到達
       [WARN] 未到達. j4 の発熱などを確認してください
[SEND] pi05_chunk006_via05: send_angles (single) [ +20.47,  +54.66,  +21.05,   -0.03,  -62.07,  -29.24]  speed=20 wait=0.0s  (pre max|Δ|=4.92°, post-send timeout 8.0s)
[SEND] [TIMEOUT] 8.0s 内に未到達: max|Δ|=6.26° actual=[ +20.21,  +55.54,  +21.26,   +0.08,  -55.81,  -28.91]
[SEND] リトライしません: pre→last max|Δ| 4.66° >= 1.0° (部分移動/ストール疑い)
       [VIA   ] step  5  Δ= +4.9  -0.9  -3.0  +0.9  -1.9  -1.4  未到達
       [WARN] 未到達. j4 の発熱などを確認してください
[SEND] pi05_chunk006_via08: send_angles (single) [ +23.77,  +54.87,  +17.37,   +1.10,  -60.30,  -30.26]  speed=20 wait=0.0s  (pre max|Δ|=4.49°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.91° (tol 3.0°)
       [VIA   ] step  8  Δ= +3.6  -0.7  -3.9  +1.0  -4.5  -1.3  着いた
[SEND] pi05_chunk006_via09: send_angles (single) [ +24.57,  +54.97,  +16.47,   +1.37,  -59.88,  -30.45]  speed=20 wait=0.0s  (pre max|Δ|=2.32°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.32° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.9  -0.8  -1.2  +0.7  -2.3  -0.9  着いた
[STATE ] 目標を渡す  目標と実測の最大差 2.32°（関節 j4）
[007] 推論 873ms  10ステップ先  Δ= +1.0  -1.4  -6.9  +1.4  -1.6  -1.0
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > r
       爪を締め直します (開かずに閉じる値 22 を送る)
  [Pi] GRIPPER: VALUE=22 speed=100 result= None
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > r
       爪を締め直します (開かずに閉じる値 25 を送る)
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > p
[PAD   ] 補正 1 (推論 7 のあと). ゲームパッドで直して, □ で推論に戻ります
コントローラーを待っています（5 秒でつながらなければ推論の入力待ちに戻ります）。□ で推論に戻ります。Ctrl-C で補正を抜けます
/opt/homebrew/Caskroom/miniforge/base/lib/python3.12/site-packages/pygame/pkgdata.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
  from pkg_resources import resource_stream, resource_exists
[接続] PS4 Controller  GUID=03008fe54c050000cc09000000016800  軸=6  ボタン=16
[送信] j1 +2.0° → +24.60 +57.98 +17.57 +1.23 -59.32 -30.14  手前で止まった（次の押しは この目標の先から）（ずれ 4.13°）  届いた角度 +24.60 +58.71 +17.40 +1.66 -63.45 -30.14
[送信] j1 +2.0° → +24.60 +59.98 +17.57 +1.23 -59.32 -30.14  届いた（ずれ 4.13°）  届いた角度 +24.60 +59.76 +16.78 +1.66 -63.45 -30.14
[送信] j1 +2.0° → +24.60 +61.98 +17.57 +1.23 -59.32 -30.14  手前で止まった（次の押しは この目標の先から）（ずれ 4.31°）  届いた角度 +24.60 +61.25 +15.90 +1.58 -63.63 -30.14
  [Pi] GRIPPER: OPEN result= None
[爪] 爪を開く（待ち 0.8 秒のあと受付）
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[爪] 爪を閉じる（待ち 0.8 秒のあと受付）
[送信] j1 -2.0° → +24.60 +59.98 +17.57 +1.23 -59.32 -30.14  手前で止まった（次の押しは この目標の先から）（ずれ 1.67°）  届いた角度 +24.60 +61.25 +15.90 +1.75 -59.06 -30.41
[長押し] j1 を +61.25° → +53.87° まで（目標を 1 回送信。止めます）
[止める] stop（released_while_sending）  止まった角度 +24.52 +53.43 +17.49 +1.84 -57.48 -30.49
[止める] stop（released_while_sending）  止まった角度 +24.52 +59.85 +17.40 +1.66 -59.50 -30.32
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[爪] 爪を閉じる（待ち 0.8 秒のあと受付）
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[爪] 爪を閉じる（待ち 0.8 秒のあと受付）
[送信] j1 -2.0° → +24.60 +57.85 +17.57 +1.23 -59.32 -30.14  手前で止まった（次の押しは この目標の先から）（ずれ 1.47°）  届いた角度 +24.52 +59.32 +17.22 +1.40 -58.88 -30.58
[送信] j1 -2.0° → +24.60 +55.85 +17.57 +1.23 -59.32 -30.14  手前で止まった（次の押しは この目標の先から）（ずれ 2.46°）  届いた角度 +24.52 +57.56 +17.22 +1.49 -56.86 -30.58
[止める] stop（released_while_sending）  止まった角度 +24.43 +53.70 +17.31 +1.66 -55.01 -30.49
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[爪] 爪を閉じる（待ち 0.8 秒のあと受付）
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[爪] 爪を閉じる（待ち 0.8 秒のあと受付）
[長押し] j1 を +53.70° → +46.66° まで（目標を 1 回送信。止めます）
[PAD   ] 補正 1 を終えました (61.6 秒, 送った目標 11 回, 爪を動かした). この姿勢から推論に戻ります
[STATE ] 実測を渡す (直前に送った目標がない)
[008] 推論 822ms  10ステップ先  Δ=+14.7 +10.8 -10.5  +1.5  +0.0  -0.6
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > h
[SEND] pi05_home: send_angles (single) [  +0.00,   +0.00,   +0.00,   +0.00,   +0.00,  -30.00]  speed=20 wait=0.0s  (pre max|Δ|=60.20°, post-send timeout 11.0s)
  [WARN] daemon_get_angles failed: daemon error: get_angles returned empty/invalid after 12 retries
  [WARN] daemon_get_angles failed: daemon error: get_angles returned empty/invalid after 12 retries
  [WARN] daemon_get_angles failed: daemon error: get_angles returned empty/invalid after 12 retries
[SEND] arrival OK: max|Δ|=1.66° (tol 2.5°)
[RECORD] 記録終了 コマ 1802 (8.888 fps) 実測角 1848 回 (9.11 Hz)  電流 1865 行 (read_servo)  → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161402/recording
       結果: y=成功 / n=失敗 / Enter=記録しない > n
[GRIP  ] 閉じて 1 秒後の爪の電流 (中央値): 1 回目 814.5 (10 行) / 2 回目 736.5 (10 行)  (閉じる値 25)
[DONE  ] 記録: /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161402
[CLIENT] run_policy_client.py が終わった (終了コード 0. モデル deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1))
    (210 秒)
もう1回やりますか? y=やる / 他=終わる > y

=== 7. HOME → 爪を開く → 位置合わせ (3 回目)
[INIT] daemon 接続中...
  [Pi] fresh_mode=1
[INIT] fresh_mode=1
[PRE]  現在姿勢: [-0.17, 1.31, 1.66, 0.26, -0.26, -30.23]
[SEND] HOME へ移動指令
[WAIT] t=0.6s  max_err=1.66°
[DONE] HOME 到達 max_err=1.66°
[POST] 最終姿勢: [-0.17, 1.31, 1.66, 0.26, -0.26, -30.23]
  [Pi] fresh_mode=0
[EXIT] fresh_mode=0, socket close
  [Pi] GRIPPER: OPEN result= None
[REF   ] 基準画像: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_194818/frames/000000.jpg
[REF   ] お手本 run_0001_20261019_194818 (つかんだ瞬間の j0 = +13.2°)
[REF   ] 重なった画像のブロックに実物のブロックを合わせると, お手本と同じ位置に置ける
[TILT  ] 傾きのお手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  中央値 -0.61° (測れた 50/50 本. プラス=右下がり)
[TILT  ] 傾きはソフトでそろえる: ライブをお手本の中央値との差だけ回して重ねる (AI に渡す画像と同じ. 回す角度が 3° を超えたら画面の色を変える. 止めない)
[ALIGN ] |dy| <= 20 なら OK. アームは HOME に置いた状態で合わせる. q または Esc で終了 (ファイルには何も書かない)
[ALIGN ] dy=  +0.2  dx=  +0.7  response=0.921  → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  傾き +0.57° → -0.61° にソフトでそろえる (-1.19° 回して重ねる. プラス=右下がり)
[DONE  ] 終了 (何も書いていません)
    (9 秒)

=== 8. 実機推論 (3 回目)
[WAIT  ] 送ったあと 0.0s / 爪のあと 0.8s (開くときの daemon 0.0s) / 止まるのを待つ目安 0.8s
[GRIP  ] 閉じる値 25  締め直し r は 22 と 25 を交互に送る
[SAFE  ] 安全範囲 (学習 50 本 ±10°): j0[-72,+55]  j1[-9,+88]  j2[-45,+52]  j3[-18,+27]  j4[-87,+10]  j5[-51,-15]
[SAFE  ] 肩 j1 の上限は J1_SAFE_MAX=88° で切った (これを超える指令はアームが黙って捨てる / LOG352)  元データ: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes
[MODE  ] EXECUTE  steps/chunk=10  max_move=20.0°  log=/Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161743
[MODEL ] deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)
[CAMERA] 写り方 dy=-5.8 dx=+7.7 response=0.561 → OK (|dy|<=20 / |dx|<=20 / response>=0.20)  回したあとの画像で測った
[CAMERA] 基準: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_teach_v1_trim2/raw_episodes/run_0001_20261019_183454/frames/000000.jpg
[CAMERA] 傾き +0.57° → -0.61° にそろえて渡す (-1.19° 回す. プラス=右下がり) (お手本: /Volumes/KIOXIA/r2r2r-data/datasets/mecharm_redblock_box_deep_even10_v1/raw_episodes  測れた 50/50 本・今 5/5 コマ)
[POLICY] 接続 ws://localhost:8000  metadata={}
[RECORD] 上カメラと実測角を 1 秒 10 コマで記録します → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161743/recording
[GRIP  ] 開始時のグリッパー開放は, 最初の目標が検査を通ってから行います (爪が閉じていたら手で開けておく)
[STATE ] 実測を渡す (直前に送った目標がない)
[000] 推論 2226ms  10ステップ先  Δ= +3.2 +12.2  +8.9  -1.4 -18.4  +2.0
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[GRIP  ] 開始時にグリッパーを開く
  [Pi] GRIPPER: OPEN result= None
[SEND] pi05_chunk000_via02: send_angles (single) [  +0.35,   +3.75,   +3.88,   +0.31,   -3.40,  -30.00]  speed=20 wait=0.0s  (pre max|Δ|=3.14°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.74° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.5  +2.4  +2.2  +0.0  -3.1  +0.2  着いた
[SEND] pi05_chunk000_via05: send_angles (single) [  +1.54,   +7.26,   +6.66,   -0.21,   -9.15,  -29.38]  speed=20 wait=0.0s  (pre max|Δ|=7.49°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.56° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.7  +2.6  +1.7  -0.5  -7.5  +0.8  着いた
[SEND] pi05_chunk000_via08: send_angles (single) [  +2.84,  +12.24,   +9.82,   -0.98,  -16.85,  -28.49]  speed=20 wait=0.0s  (pre max|Δ|=9.21°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.47° (tol 3.0°)
       [VIA   ] step  8  Δ= +1.9  +4.1  +2.3  -1.0  -9.2  +1.1  着いた
[SEND] pi05_chunk000_via09: send_angles (single) [  +3.03,  +13.51,  +10.55,   -1.14,  -18.67,  -28.28]  speed=20 wait=0.0s  (pre max|Δ|=3.29°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.01° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.4  +0.2  -0.2  +0.0  -3.3  +0.5  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.01°（関節 j4）
[001] 推論 869ms  10ステップ先  Δ= +5.3 +12.5  +8.1  -1.1 -23.0  +1.5
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk001_via02: send_angles (single) [  +4.21,  +16.92,  +13.02,   -1.89,  -24.48,  -27.60]  speed=20 wait=0.0s  (pre max|Δ|=6.82°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.11° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.6  +2.6  +1.8  -0.8  -6.8  +0.8  着いた
[SEND] pi05_chunk001_via05: send_angles (single) [  +5.90,  +21.18,  +15.92,   -2.21,  -31.66,  -27.17]  speed=20 wait=0.0s  (pre max|Δ|=8.20°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.99° (tol 3.0°)
       [VIA   ] step  5  Δ= +2.0  +3.3  +2.0  -0.2  -8.3  +0.7  着いた
[SEND] pi05_chunk001_via08: send_angles (single) [  +7.57,  +25.71,  +18.86,   -2.18,  -38.96,  -26.93]  speed=20 wait=0.0s  (pre max|Δ|=8.29°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.92° (tol 3.0°)
       [VIA   ] step  8  Δ= +2.1  +3.6  +2.2  +0.3  -8.3  +0.5  着いた
[SEND] pi05_chunk001_via09: send_angles (single) [  +7.92,  +26.86,  +19.38,   -2.21,  -40.63,  -26.89]  speed=20 wait=0.0s  (pre max|Δ|=2.49°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.91° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.9  +0.2  -0.1  +0.5  -2.5  +0.3  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.91°（関節 j1）
[002] 推論 825ms  10ステップ先  Δ= +3.0 +11.6  +6.0  +0.8 -19.0  +0.4
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk002_via02: send_angles (single) [  +9.16,  +32.05,  +21.92,   -2.24,  -48.41,  -26.83]  speed=20 wait=0.0s  (pre max|Δ|=8.07°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.78° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.5  +4.3  +2.1  +0.4  -8.1  +0.3  着いた
[SEND] pi05_chunk002_via05: send_angles (single) [ +10.18,  +36.27,  +24.02,   -1.98,  -54.66,  -26.81]  speed=20 wait=0.0s  (pre max|Δ|=7.03°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.55° (tol 3.0°)
       [VIA   ] step  5  Δ= +1.2  +3.3  +1.5  +0.7  -7.0  +0.3  着いた
[SEND] pi05_chunk002_via08: send_angles (single) [ +10.54,  +38.94,  +25.56,   -1.87,  -58.63,  -26.78]  speed=20 wait=0.0s  (pre max|Δ|=4.58°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.69° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.3  +1.8  +1.0  +0.7  -4.6  +0.3  着いた
[SEND] pi05_chunk002_via09: send_angles (single) [ +10.64,  +39.38,  +25.87,   -1.85,  -59.31,  -26.78]  speed=20 wait=0.0s  (pre max|Δ|=0.87°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.78° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.6  -0.4  -0.1  +0.6  -1.0  +0.1  着いた
[STATE ] 目標を渡す  目標と実測の最大差 0.78°（関節 j1）

[003] 推論 905ms  10ステップ先  Δ= +2.7  +6.4  +2.7  +0.2  -6.8  +0.1
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk003_via02: send_angles (single) [ +11.81,  +43.28,  +28.40,   -2.13,  -64.14,  -26.79]  speed=20 wait=0.0s  (pre max|Δ|=5.35°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.92° (tol 3.0°)
       [VIA   ] step  2  Δ= +1.8  +3.1  +2.5  +0.2  -5.3  +0.1  着いた
[SEND] pi05_chunk003_via05: send_angles (single) [ +12.23,  +44.97,  +29.18,   -2.14,  -65.28,  -26.80]  speed=20 wait=0.0s  (pre max|Δ|=1.74°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.90° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.5  +0.8  +0.5  +0.0  -1.7  -0.0  着いた
[SEND] pi05_chunk003_via08: send_angles (single) [ +12.52,  +45.78,  +28.91,   -2.16,  -65.45,  -26.81]  speed=20 wait=0.0s  (pre max|Δ|=0.84°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=0.97° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.8  -0.1  -0.2  +0.0  -0.2  -0.0  着いた
       [VIA   ] step  9  Δ= +0.1  -0.2  -0.4  -0.1  -0.4  -0.0  送らず (0.5° 以下)
[STATE ] 目標を渡す  目標と実測の最大差 0.42°（関節 j2）
[004] 推論 808ms  10ステップ先  Δ= +0.2 +14.3 -10.1  -0.1  +1.9  -0.5
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk004_via02: send_angles (single) [ +12.89,  +50.34,  +25.99,   -2.17,  -64.56,  -26.85]  speed=20 wait=0.0s  (pre max|Δ|=3.59°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.78° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.3  +3.6  -3.1  -0.2  +0.7  -0.0  着いた
[SEND] pi05_chunk004_via05: send_angles (single) [ +12.96,  +55.87,  +22.09,   -2.10,  -63.60,  -26.99]  speed=20 wait=0.0s  (pre max|Δ|=4.98°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.90° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.4  +4.6  -5.0  -0.1  +1.6  -0.2  着いた
[SEND] pi05_chunk004_via08: send_angles (single) [ +12.92,  +60.10,  +19.39,   -2.14,  -63.40,  -27.16]  speed=20 wait=0.0s  (pre max|Δ|=3.72°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.81° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.4  +3.4  -3.7  -0.1  +1.8  -0.4  着いた
[SEND] pi05_chunk004_via09: send_angles (single) [ +12.72,  +61.07,  +18.94,   -2.12,  -63.33,  -27.26]  speed=20 wait=0.0s  (pre max|Δ|=1.88°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.88° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.2  +0.1  -1.4  -0.1  +1.9  -0.6  着いた
[STATE ] 目標を渡す  目標と実測の最大差 1.88°（関節 j4）
[005] 推論 808ms  3ステップ先  Δ= +0.3  -0.5  -1.2  -0.0  +1.8  -0.3  グリッパー→閉
       [VIA   ] 中継点 1 点: step 2
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk005_via02: send_angles (single) [ +12.86,  +61.22,  +18.70,   -2.12,  -63.45,  -27.33]  speed=20 wait=0.0s  (pre max|Δ|=1.67°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.67° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.3  -0.5  -1.2  -0.0  +1.7  -0.3  着いた
  [Pi] GRIPPER: VALUE=25 speed=100 result= None
[STATE ] 目標を渡す  目標と実測の最大差 2.11°（関節 j4）
[006] 推論 2178ms  10ステップ先  Δ= +1.1 -10.2  +8.9  -1.2  +3.3  +0.4
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > 
[SEND] pi05_chunk006_via02: send_angles (single) [ +13.06,  +58.14,  +21.19,   -2.37,  -61.69,  -27.14]  speed=20 wait=0.0s  (pre max|Δ|=3.70°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.09° (tol 3.0°)
       [VIA   ] step  2  Δ= +0.8  -3.6  +2.0  -0.8  +3.7  +0.4  着いた
[SEND] pi05_chunk006_via05: send_angles (single) [ +13.15,  +53.54,  +25.65,   -2.62,  -60.51,  -26.52]  speed=20 wait=0.0s  (pre max|Δ|=5.61°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.04° (tol 3.0°)
       [VIA   ] step  5  Δ= +0.9  -5.6  +5.4  -0.8  +1.2  +1.1  着いた
[SEND] pi05_chunk006_via08: send_angles (single) [ +13.26,  +51.60,  +28.31,   -2.56,  -61.72,  -26.42]  speed=20 wait=0.0s  (pre max|Δ|=3.53°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=1.96° (tol 3.0°)
       [VIA   ] step  8  Δ= +0.7  -3.0  +3.5  +0.3  -2.1  +0.6  着いた
[SEND] pi05_chunk006_via09: send_angles (single) [ +13.40,  +51.58,  +28.27,   -2.45,  -62.23,  -26.56]  speed=20 wait=0.0s  (pre max|Δ|=2.29°, post-send timeout 8.0s)
[SEND] arrival OK: max|Δ|=2.12° (tol 3.0°)
       [VIA   ] step  9  Δ= +0.6  -1.1  +0.6  +0.5  -2.3  +0.2  着いた
[STATE ] 目標を渡す  目標と実測の最大差 2.12°（関節 j4）
[007] 推論 860ms  10ステップ先  Δ=+15.1  -0.6 -13.5  +3.3  +2.5  -3.8
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > p
[PAD   ] 補正 1 (推論 7 のあと). ゲームパッドで直して, □ で推論に戻ります
コントローラーを待っています（5 秒でつながらなければ推論の入力待ちに戻ります）。□ で推論に戻ります。Ctrl-C で補正を抜けます
/opt/homebrew/Caskroom/miniforge/base/lib/python3.12/site-packages/pygame/pkgdata.py:25: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
  from pkg_resources import resource_stream, resource_exists
[接続] PS4 Controller  GUID=03008fe54c050000cc09000000016800  軸=6  ボタン=16
[長押し] j1 を +52.64° → +52.64° まで（目標を 1 回送信。止めます）
[PAD   ] 補正 1 を終えました (15.3 秒, 送った目標 2 回). この姿勢から推論に戻ります
[STATE ] 実測を渡す (直前に送った目標がない)
[008] 推論 2257ms  10ステップ先  Δ= +0.7  +0.6  -0.5  +0.5  -4.6  -0.2
       [VIA   ] 中継点 4 点: step 2, 5, 8, 9
       Enter=実行 / s=送らず次へ / b=巻き戻し / p=ゲームパッドで補正 / c=キーで補正 / r=爪の締め直し / h=HOMEに戻して終了 / q=終了 > h
[SEND] pi05_home: send_angles (single) [  +0.00,   +0.00,   +0.00,   +0.00,   +0.00,  -30.00]  speed=20 wait=0.0s  (pre max|Δ|=63.28°, post-send timeout 11.5s)
[SEND] arrival OK: max|Δ|=2.46° (tol 2.5°)
[RECORD] 記録終了 コマ 860 (8.969 fps) 実測角 891 回 (9.29 Hz)  電流 900 行 (read_servo)  → /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161743/recording
       結果: y=成功 / n=失敗 / Enter=記録しない > n
[GRIP  ] 閉じて 1 秒後の爪の電流 (中央値): 1 回目 703 (11 行)  (閉じる値 25)
[DONE  ] 記録: /Volumes/KIOXIA/r2r2r-data/runs/pi05_policy/pi05_20261111_161743
[CLIENT] run_policy_client.py が終わった (終了コード 0. モデル deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1))
    (8173 秒)
もう1回やりますか? y=やる / 他=終わる > 

=== かかった時間  モデル deep1 (PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)
  1. Pod を作る                      4 秒
  2. Pod が立ち上がるまで待つ     40 秒
  3a. スクリプトを送る        122 秒
  3b. huggingface_hub を入れ, 鍵を渡す     14 秒
  3c. モデルを取る (hf download PixelstoMotors/pi05-mecharm-deep-even10-lora-v1)     14 秒
  3d. openpi を取る (215abfb)       22 秒
  3e. uv sync                          21 秒
  3f. addon・設定・assets (pi05_mecharm_deep_even10_lora)      3 秒
  4. サーバーの起動             53 秒
  5. Mac の 8000 番へのトンネル      3 秒
  6. デーモンの状態              2 秒
  7. HOME → 爪を開く → 位置合わせ (1 回目)    170 秒
  8. 実機推論 (1 回目)          300 秒
  7. HOME → 爪を開く → 位置合わせ (2 回目)     10 秒
  8. 実機推論 (2 回目)          210 秒
  7. HOME → 爪を開く → 位置合わせ (3 回目)      9 秒
  8. 実機推論 (3 回目)         8173 秒

[LAB   ] Pod は残しています (id 1zx0vdv9xasenp). 片付けは ./lab_stop.sh --log Log30
[TUNNEL] 8000 番のトンネルを閉じました
