CoolFace
Datasetpublic

THULab/eval3_90_permutation

eval3_90_permutation (TsFile) Apache TsFile version of robot-learning-group47/eval3_90_permutation. Overview A LeRobot (v3.0) robot-manipulation evaluation dataset collected on a so_follower arm. Each episode records a pick-and-place rollout — placing a coke can onto one of three printed portraits — sampled frame by frame as commanded action and measured arm state. Robot: so_follower (6-DoF arm). Tasks (3): "Place the coke on Taylor Swift." "Place the coke on… See the full description on the dataset page: https://huggingface.co/datasets/THULab/eval3_90_permutation.

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
0likes47downloads
README.md83 linesDownload Raw Back to root
1---2license: apache-2.03task_categories:4- robotics5tags:6- LeRobot7- robotics8- tsfile9- modality:timeseries10- timeseries11- format:tsfile12pretty_name: eval3_90_permutation (TsFile)13size_categories:14- 10K<n<100K15configs:16- config_name: default17  data_files:18  - split: train19    path: data/eval3_90_permutation.tsfile20modality:21- tabular22- timeseries23---24 25# eval3_90_permutation (TsFile)26 27Apache TsFile version of [`robot-learning-group47/eval3_90_permutation`](https://huggingface.co/datasets/robot-learning-group47/eval3_90_permutation).28 29## Overview30 31A [LeRobot](https://github.com/huggingface/lerobot) (v3.0) robot-manipulation32evaluation dataset collected on a `so_follower` arm. Each episode records a33pick-and-place rollout — placing a coke can onto one of three printed portraits —34sampled frame by frame as commanded action and measured arm state.35 36- **Robot:** `so_follower` (6-DoF arm).37- **Tasks (3):**38  - "Place the coke on Taylor Swift."39  - "Place the coke on Barack Obama."40  - "Place the coke on Yann LeCun."41- **Scale:** 90 episodes, 20,221 frames, single `train` split.42- **Sampling rate:** 15 fps.43- **Videos:** one front camera stream (`observation.images.front`) in the original44  dataset; not included here (see below).45 46## Schema (TsFile structure)47 48- **Time** (INT64, milliseconds) — `round(timestamp * 1000)`, restarts per episode.49- **episode_index** (TAG) — episode device dimension; query one episode with50  `WHERE episode_index=0`.51- **task_index** (TAG) — task device dimension (3 tasks: 0/1/2 as listed above).52- **frame_index** (FIELD, INT64) — original per-episode frame counter.53- **sample_index** (FIELD, INT64) — source `index` column, renamed.54- **action_0..action_5** (FIELD, FLOAT) — commanded 6-DoF action.55- **observation_state_0..observation_state_5** (FIELD, FLOAT) — measured 6-DoF56  arm state.57 58Vector columns are flattened by preserving the source name (`.` → `_`) and59appending the element index. The source `timestamp` column is dropped because it60equals `Time ÷ 1000` seconds.61 62## Attachments63 64- `eval3_episode_permutation_map.csv` / `eval3_episode_permutation_map.json` —65  the evaluation episode permutation map, mirrored as-is from the source dataset.66 67## Videos68 69The front camera video stream is not included in this TsFile repository. It70remains in the original dataset under71[`videos/`](https://huggingface.co/datasets/robot-learning-group47/eval3_90_permutation/tree/main/videos).72 73## Usage74 75Read the `.tsfile` files with the Apache TsFile Java or Python SDK.76 77## Source & license78 79- Original dataset: https://huggingface.co/datasets/robot-learning-group47/eval3_90_permutation80- Author / publisher: robot-learning-group4781- Created with: [LeRobot](https://github.com/huggingface/lerobot) (codebase v3.0)82- License: apache-2.083