WMX R2 CR3A — axis_1000hz
Public cumulative release: 204 recorded runs = the original 59 + 145 audited additions. LeRobot v3.0; 3,360,238 frames; nominal 1000 Hz. These runs are not 204 confirmed successful tasks.
Axis 데이터 구성과 시간축
이 데이터셋은 WMX3의 EtherCAT DC 동기화된 1 kHz 제어 주기에서 취득한 축 명령·피드백 데이터입니다. 고주파 축 동작과 추종 오차 분석을 위해 204개 실행의 원본 축 로그 3,360,238행과 원본 사이클 번호를 모두 유지합니다. DC 동기화된 제어 주기에서의 취득과, 아래에 정의한 LeRobot 상대 시간 표현을 구분합니다.
| 항목 | 정의 |
|---|---|
| 취득 주기 | WMX3 EtherCAT DC 동기화된 1 kHz 제어 주기, 명목 1 ms |
| LeRobot timestamp | 에피소드 시작을 0초로 하는 상대 시간: frame_index / 1000, 초 단위 |
| source.cycle | 원본 엔진 사이클 번호 CYCLE |
| source.axis_row | 원본 축 로그의 행 번호 |
| source.axis_host_time_ns | 원본 SYSTEM_TIME × 100, 정수 나노초 호스트 시각 |
| 수치 정밀도 | 학습 특성 float32, 원본 정밀도는 raw 패키지에 보존 |
축 파생본은 원본 축 로그를 다운샘플링하지 않습니다. 다만 학습용 명령·피드백 값은 float32로 변환되므로 원본 파일과 비트 단위로 동일한 표현은 아닙니다. 원본 CSV와 원본 정밀도의 Parquet는 raw 패키지에 보존합니다.
LeRobot의 timestamp는 변환 시 생성한 상대 시간이며, 에피소드마다 0초에서 시작합니다. 원본 DC 절대 시각을 복사한 열이 아닙니다. source.cycle은 원본 CYCLE, source.axis_row는 원본 행 번호, source.axis_host_time_ns는 원본 SYSTEM_TIME × 100으로 변환한 정수 나노초 호스트 시각입니다. 원본 TIME 및 SYSTEM_TIME은 raw 패키지에 그대로 남아 있습니다. 균일한 timestamp만으로 실제 시간 간격의 변동을 평가하지 말고 원본 시간 열을 함께 확인해야 합니다.
DC 동기화된 1 kHz 제어 주기에서 취득했다는 점은 데이터 취득 구성에 대한 설명입니다. LeRobot timestamp = frame_index / 1000은 그 샘플 순서를 1 ms 간격의 상대 시간축으로 표현한 값이며, DC 절대 시각을 직접 저장한 값은 아닙니다. SYSTEM_TIME은 별도로 보존한 호스트 시각으로, DC 시계와 동일한 원본 시간값이라고 해석하지 않습니다. 이번 패키징 검증은 원본 사이클 연속성과 데이터 대응을 확인한 것이며, DC 동기화의 오차를 별도로 실측한 검증은 아닙니다.
English summary
This dataset contains axis commands and feedback acquired during WMX3 EtherCAT DC-synchronized 1 kHz control cycles. Every original axis-log row and cycle number is retained. LeRobot timestamp = frame_index / 1000 represents relative elapsed time in seconds, starting at zero for each episode with nominal 1 ms spacing; it is not a directly recorded absolute DC clock value. Acquisition synchronization and timestamp representation are distinct. Engine cycles, source row indices and host timestamps remain separate int64 provenance fields; raw TIME and SYSTEM_TIME remain in the original archive. SYSTEM_TIME is retained as host time and is not asserted to be the DC clock value. Model signals are cast to float32. Packaging checks establish cycle continuity and source-data correspondence, not an independent measurement of DC synchronization error.
Loading
Tested with Python 3.12, lerobot==0.4.4 and av==15.1.0; see requirements-lock.txt for the complete environment.
from lerobot.datasets.lerobot_dataset import LeRobotDataset
ds = LeRobotDataset("movensys/wmx-r2-command-grounded-cr3a-axis-v0", revision="cumulative-204-language-v1-20261102", video_backend="pyav")
sample = ds[0]
Initial download requires network access. No robot commands are issued. Keep meta/ and data/ together. For a local download, pass root="/absolute/path/to/download".
Original data, labels and examples
Original CSV/AVI, timestamps and raw examples are stored once in the axis repository raw package. Raw package paths in annotations resolve relative to that directory. annotations/ contains integrated labels and episode_mapping.json; original source paths are provenance identifiers. Use examples/plot_dataset.py with --root pointing to this downloaded repository and --episode 162 --output example.png. plots/ contains executed examples. The bundled load_local.py and validate_bundle.py target the full local bundle layout; the direct loading example above targets this standalone repository.
Feature meanings
action: six command positions then six command velocities (configured rad, rad/s). observation.state: six feedback positions then six feedback velocities. observation.tracking_error: command-minus-feedback positions computed before float32 conversion. Engine setpoints are not a complete deployed policy interface. Source indices and timestamps remain int64; raw numeric data stays float64. FBTRQ is excluded from learning features because ENI mapping is unverified. No measured gripper force/position or continuous depth is claimed.
Validation and quality
All 3,360,238 numeric rows were compared with the source at the declared float32 precision. Original cycle continuity and increasing host timestamps were checked. First, middle and last samples of all 204 episodes were loaded with the official LeRobot loader, and batch loading was verified. See loading_validation.json. The source was acquired in DC-synchronized 1 kHz control cycles; the generated relative timeline is not an independent measurement of DC synchronization accuracy. Engine setpoints and feedback do not establish physical task success.
Labels and scope
The baseline59 remains a fixed cohort. Additional T008 correction runs are now included separately and do not overwrite T007's partial orientation outcome. Incoming corrections and evidence addenda are captured in the integrated label overlay. Unknown physical outcomes remain unreviewed; controller success, user confirmation and image estimates are separate. Missing task links remain null; UNASSIGNED per-run lookup keys are placeholders, not new confirmed task chains. Consult annotations/integration_notes.json and episode applied_addenda. Labels are not a newly adjudicated physical-success ground truth. Original source ledgers were not changed.
No held-out split is supplied. Group related work by task/session; the default LeRobot train split is storage only. Original source_path strings remain provenance identifiers; use package paths and episode_mapping for loading. Failure/multistream cases not selected for the 145 additions remain outside this converted cohort.
Versions and license
This release replaces the default 59-run snapshot with the cumulative 204-run snapshot; do not append all 204 to the old 59. The former version is preserved as baseline-59-before-20261102. Pin revision="cumulative-204-language-v1-20261102" for reproducibility. Aggregate statistics were recomputed over 204 runs. No explicit open-data license is assigned; public visibility does not itself grant an open license.
Language instruction correction on 2026-11-02
135 added episodes (134 distinct task strings) previously used task IDs as their descriptive prefix. They now have Korean descriptions of intended operations in both meta/tasks.parquet and meta/episodes. These descriptions are based on the recorded operations, not new physical-success judgments. The original 59 episode instructions, frame task_index values, numeric data, videos, task identities and 35 null task links are unchanged. annotations/language_instruction_overrides.json records the before/after wording and evidence basis. The prior cumulative-204-20261102 tag remains immutable. See language_instruction_validation.json for per-episode official-loader checks.
- Downloads last month
- 261