Apache TsFile version of justintiensmith/rolloutpi05ReasoningStep076596Counting. A SO follower rollout dataset from a pi0.5 policy on a counting task family. The 32 language instructions ask the robot to move objects (a blue pen, a green pencil sharpener, a yellow block and an airplane toy) into or out of a bowl so that the bowl ends up holding a requested number of objects. Each frame records the 6-D joint state and the 6-D commanded action. Tasks (taskindex 0-31) include instructions such as "Leave exactly one object in the bowl", "Adjust the bowl so that it contains two objects", "Leave exactly three objects in the bowl", "Adjust the bowl so that it contains zero objects", "Move the…
Publicly accessible
apache-2.0
100K<n<1M
Apache TsFile version of weiye11/0603pi0onlyright. A right-arm-only teleoperation dataset recorded on an SO-100 robot and packaged in the LeRobot v2.1 layout. Each row is one control step: a 6-dimensional right-arm joint/gripper state and the 6-dimensional commanded action. The source dataset card does not name the manipulation task; the metadata records a single task (taskindex = 0). - Time (INT64, milliseconds) — round(timestamp 1000); restarts at 0 for each episode. - episodeindex (TAG) — episode / device dimension. - taskindex (TAG) — task dimension (constant 0 here). - frameindex (FIELD, INT64) — source frame counter; sampleindex (FIELD, INT64) — source index column (renamed).…
Publicly accessible
apache-2.0
1K<n<10K
Apache TsFile version of Shiki42/PutCab-AB-RandomDelta-Train50. A timing-balanced 50-episode subset of RoboTwin PutCab-AB demonstrations, recorded on a simulated ALOHA bimanual robot. Selection is conditioned on the normalized reference offset uref = (delta + TBref) / (TAref + TBref) (where delta is right/B start minus left/A start): ten intervals 0,.1)... [.9,1) each contribute five trajectories, giving 25 negative and 25 positive deltas. Reused trajectories keep independent active-arm v3 timing; the new V2 trajectories add recorded feedback waits (safe-waypoint lift, drawer-readiness re-check). observation.state holds 14 joint/gripper drive targets and action is the next observation…
Publicly accessible
10K<n<100K
Apache TsFile version of zwonzis/xarm7-pick-orange-ball-large. A UFactory xArm7 dataset for the task "pick up the orange ball and put it in the cyan bowl". Each frame records the 8-D arm state and the 8-D commanded action. Task (taskindex 0): "pick up the orange ball and put it in the cyan bowl". - Time (INT64, milliseconds) - per-episode sample time, round(timestamp 1000) in milliseconds (the source timestamp column equals Time / 1000 seconds and is therefore not stored separately). - episodeindex, taskindex (TAG) - device dimensions identifying one demonstration episode and one task; query a single episode with WHERE episodeindex=0. - observationstate0..7, action0..7 (FIELD) - FLOAT 8-D…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of axiboai/piper-umi-ee-pick-place. This is a pick-and-place demonstration dataset recorded on an AgileX Piper arm with a single 6-DoF end-effector (pipersingleeef6d) and packaged in the LeRobot v2.1 layout. Each row is one control step: a 10-dimensional end-effector state (position x, y, z, rotation as 6-D rot6d, and gripper) together with the 10-dimensional commanded action. The metadata records a single task (taskindex = 0). - Time (INT64, milliseconds) — round(timestamp 1000); restarts at 0 for each episode. - episodeindex (TAG) — episode / device dimension. - taskindex (TAG) — task dimension (constant 0 here). - frameindex (FIELD, INT64) — source frame counter…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of nduque/evalrobustnesse93fullfklocal240k. This is an evaluation-robustness teleoperation dataset recorded on a Koch robot arm and packaged in the LeRobot v2.1 layout. Each row is one control step: a 6-dimensional joint/gripper state, the 6-dimensional commanded action, and a 4-dimensional local end-effector pose. The source dataset card does not name the manipulation task; the metadata records a single task (taskindex = 0). - Time (INT64, milliseconds) — round(timestamp 1000); restarts at 0 for each episode. - episodeindex (TAG) — episode / device dimension. - taskindex (TAG) — task dimension (constant 0 here). - frameindex (FIELD, INT64) — source frame counter…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of kjydb/cafe001. A teleoperation dataset recorded on a PS-A-03 500 arm and packaged in the LeRobot v2.1 layout. Each row is one control step: a 13-dimensional Cartesian state (position x, y, z, orientation rx, ry, rz, linear velocities x.vel, y.vel, z.vel, angular velocities pitch.vel, roll.vel, yaw.vel, and gripper.status) and the matching 13-dimensional commanded action. The metadata records a single task (taskindex = 0). - Time (INT64, milliseconds) — round(timestamp 1000); restarts at 0 for each episode. - episodeindex (TAG) — episode / device dimension. - taskindex (TAG) — task dimension (constant 0 here). - frameindex (FIELD, INT64) — source frame counter…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of MA-tokenweights/all-the-news-2-pythia-tfidf-topic-stratified-v1-articles. The articles/text table of the MA-tokenweights All the News 2 Pythia TF-IDF topic-stratified pipeline. Each record is one news article drawn from the All the News 2 corpus: its full body text, word/token counts and line spans, plus bibliographic metadata (title, publication, date, URL, section). The corpus was tokenized with the Pythia tokenizer and weighted with TF-IDF, topic-stratified, and split into trainarticles and validationarticles. (163,380 articles in total). (metatitle, metapublication, metadate, metaurl, metasection,...). articles came from the original training split); the two…
Publicly accessible
100K<n<1M
Apache TsFile version of nikodembartnik/so101pencilholder1. A SO follower dataset for the task "put the pen in the wooden box" (source pretty name: "Task: put pen in wooden box"). Each frame records the 6-D joint state and the 6-D commanded action. Task (taskindex 0): "put the pen in the wooden box". - Time (INT64, milliseconds) - per-episode sample time, round(timestamp 1000) in milliseconds (the source timestamp column equals Time / 1000 seconds and is therefore not stored separately). - episodeindex, taskindex (TAG) - device dimensions identifying one demonstration episode and one task; query a single episode with WHERE episodeindex=0. - action0..5, observationstate0..5 (FIELD) - FLOAT…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of nitsuga-hug/augs-hiliter01. A SO follower (SO-101/SO-100) dataset for the "pnphiliter" pick-and-place task, collected with augmented human-in-the-loop data. Each frame records the 6-D joint state and the 6-D commanded action. Task (taskindex 0): "pnphiliter". - Time (INT64, milliseconds) - per-episode sample time, round(timestamp 1000) in milliseconds (the source timestamp column equals Time / 1000 seconds and is therefore not stored separately). - episodeindex, taskindex (TAG) - device dimensions identifying one demonstration episode and one task; query a single episode with WHERE episodeindex=0. - action0..5, observationstate0..5 (FIELD) - FLOAT 6-D SO-follower…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of argus-systems/pickup-carrot-remove-parquet-metadata-2. A Trossen AI stationary-arm dataset for a single pick-and-place task: pick up a carrot and put it in the bowl. Each frame records the 14-D dual-arm action and 14-D observation state. Task (taskindex 0): "Pick up carrot and put it in the bowl". Note: the source meta/info.json and meta/episodes.jsonl declare 61 episodes / 27,261 frames, but the source repository only contains 58 data/chunk-000/episode.parquet files (episodes 0, 39 and 50 are absent); the 58 present episodes (25,922 frames) were converted. - Time (INT64, milliseconds) - per-episode sample time, round(timestamp 1000) in milliseconds (the source…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of binhng/liberoobjectmaskdepthIPECCOMMUNITYformat. A Franka robot dataset for the LIBERO-Object manipulation suite, distributed in the IPEC community format with object-mask and depth observations. Each of the ten tasks asks the robot to pick up a household object and place it in the basket. The converted time series records the robot's end-effector and joint state together with the commanded action at every control step. Tasks (taskindex): 0 pick up the salad dressing and place it in the basket; 1 bbq sauce; 2 ketchup; 3 tomato sauce; 4 alphabet soup; 5 cream cheese; 6 chocolate pudding; 7 butter; 8 orange juice; 9 milk. - Time (INT64, milliseconds) - per-episode…
Publicly accessible
apache-2.0
10K<n<100K
Apache TsFile version of taku-y/record-20260429-2. A SO-101 follower dataset for the task "Grab the toy 2". Each frame records the 6-D joint state and the 6-D commanded action. Task (taskindex 0): "Grab the toy 2". - Time (INT64, milliseconds) - per-episode sample time, round(timestamp 1000) in milliseconds (the source timestamp column equals Time / 1000 seconds and is therefore not stored separately). - episodeindex, taskindex (TAG) - device dimensions identifying one demonstration episode and one task; query a single episode with WHERE episodeindex=0. - action0..5, observationstate0..5 (FIELD) - FLOAT 6-D SO-101 joint action and 6-D observation state. - frameindex, sampleindex (INT64…
Publicly accessible
apache-2.0
10K<n<100K
