Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot extract the features (columns) for the split 'train' of the config 'default' of the dataset.
Error code:   FeaturesError
Exception:    ArrowInvalid
Message:      JSON parse error: Invalid value. in row 0
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 354, in _generate_tables
                  df = pandas_read_json(f)
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 38, in pandas_read_json
                  return pd.read_json(path_or_buf, **kwargs)
                         ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 815, in read_json
                  return json_reader.read()
                         ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1014, in read
                  obj = self._get_object_parser(self.data)
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1040, in _get_object_parser
                  obj = FrameParser(json, **kwargs).parse()
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1176, in parse
                  self._parse()
                  ~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1392, in _parse
                  ujson_loads(json, precise_float=self.precise_float), dtype=None
                  ~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
              ValueError: Expected object or value
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 244, in compute_first_rows_from_streaming_response
                  iterable_dataset = iterable_dataset._resolve_features()
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 4675, in _resolve_features
                  features = _infer_features_from_batch(self.with_format(None)._head())
                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2885, in _head
                  return next(iter(self.iter(batch_size=n)))
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3095, in iter
                  for key, pa_table in ex_iterable.iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2600, in _iter_arrow
                  yield from self.ex_iterable._iter_arrow()
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 577, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 459, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 357, in _generate_tables
                  raise e
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 295, in _generate_tables
                  pa_table = paj.read_json(
                      io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
                  )
                File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                  return check_status(status)
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: JSON parse error: Invalid value. in row 0

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Starpilot mobile UMI samples

14 samples from field takes of the Starpilot mobile UMI: one person in a real home, wearing an untethered, body-worn rig. The rig has a head stereo camera with an IMU, and two handheld grippers, each with a wrist camera, an IMU and an ArUco cube. Each sample is a complete LeRobot v3.0 dataset (one episode), plus the raw IMU, trajectories, calibration, sync and logs of the take. Explore them in the viewer.

All 4 video streams are a true 30 fps (head stereo left + right eye, left + right wrist; 1920x1200 H.264). Every stream has a distinct frame for every data row; each sample's logs/processing/frame_check.json has the check. Recorded on umi_firmware 0.9.1 to 0.16.5, 2026-11-04 to 2026-11-12.

Fixes applied after export (2026-11-12, listed in each sample's README and meta/info.json -> sample.fixes): gripper pose glitches (one-frame jumps of > 4 cm that the gripper's own IMU did not see, 50 events over the 4 samples) are marked invalid in observation.state_valid / valid.frame (values kept), and each hand's jaw opening is re-zeroed on its closed level (the left jaw read about -1 mm closed). The version straight from the exporter is the git tag as-exported.

Updated 2026-11-20: idle time removed (both grippers still and the jaws not moving for 8 s or more inside a sample, 3 s or more at its ends; active pieces under 2 minutes dropped; some samples are split into _b pieces), and every camera has a local-contrast look (fixed settings per camera and sample, in each sample's README and meta/info.json -> sample.tone). 14 samples, 85 minutes in total. Earlier versions: git tags before-local-contrast, before-idle-cut.

Sample Task Length One file
mumi_bedroom_nightstand/ Tidy the nightstand, plug in and set down a phone (bedroom nightstand) 12:11 mumi_bedroom_nightstand.zip (4.3 GB)
mumi_bedroom_nightstand_2/ Tidy the nightstand (earlier stretch) (bedroom nightstand) 9:50 mumi_bedroom_nightstand_2.zip (3.5 GB)
mumi_bedroom_nightstand_3/ Tidy the nightstand (first stretch) (bedroom nightstand) 4:41 mumi_bedroom_nightstand_3.zip (1.6 GB)
mumi_bedroom_nightstand_b/ Tidy the nightstand, plug in and set down a phone (part 2) (bedroom nightstand) 2:49 mumi_bedroom_nightstand_b.zip (1.0 GB)
mumi_hardware_sort/ Sort hardware (screws, nuts) into an organizer (bedroom) 5:17 mumi_hardware_sort.zip (1.8 GB)
mumi_hardware_sort_2/ Sort hardware into an organizer (later stretch) (bedroom) 7:15 mumi_hardware_sort_2.zip (2.3 GB)
mumi_hardware_sort_3/ Sort hardware into an organizer (middle stretch) (bedroom) 2:16 mumi_hardware_sort_3.zip (0.8 GB)
mumi_hardware_sort_3_b/ Sort hardware into an organizer (middle stretch) (part 2) (bedroom) 4:58 mumi_hardware_sort_3_b.zip (1.7 GB)
mumi_hardware_sort_b/ Sort hardware (screws, nuts) into an organizer (part 2) (bedroom) 7:13 mumi_hardware_sort_b.zip (2.4 GB)
mumi_desk_picture_shelf/ Hang a picture, position and install a shelf board (study desk) 4:37 mumi_desk_picture_shelf.zip (1.6 GB)
mumi_desk_picture_shelf_2/ Hang a picture, position a shelf board (earlier stretch) (study desk) 4:33 mumi_desk_picture_shelf_2.zip (1.6 GB)
mumi_desk_picture_shelf_b/ Hang a picture, position and install a shelf board (part 2) (study desk) 3:05 mumi_desk_picture_shelf_b.zip (1.1 GB)
mumi_desk_sort_mail/ Sort mail and papers into folders, reposition the desk (study desk) 11:58 mumi_desk_sort_mail.zip (4.7 GB)
mumi_dishwasher/ Load the dishwasher (kitchen) 3:53 mumi_dishwasher.zip (1.6 GB)
hf download starpilot-ai/mobile_umi_samples --repo-type dataset --include "mumi_desk_picture_shelf/*" --local-dir .

What is in a sample

Each sample folder has its own README.md and a manifest.json that lists every file with its size and sha256.

Required Where
Video videos/: the 4 streams of the 3 cameras, 1920x1200 H.264 at 30 fps (head stereo pair, 175° fisheye; both wrist cameras). One video frame per data row.
Task annotation meta/tasks.parquet, meta/episodes/
Post-calculation / IMU data imu/{head,left_hand,right_hand}.parquet: raw gyro + accelerometer of all three cameras (~300 Hz) on the dataset clock. data/ also has *_valid masks, valid.frame (the training filter) and qc flags.
Motion trajectories data/: observation.state / action (20-D, [right 10, left 10]): xyz (m), rot6d (the first two rows of world_from_gripper), jaw opening (m). trajectories/head.parquet: the head camera's 6-DoF path.
Metadata meta/info.json (layout, units, provenance; sample describes the window and the world frame), stats.json, modality.json
Data manifest / inventory manifest.json, README.md
Calibration and synchronization calibration/: all camera intrinsics/extrinsics, head camera-IMU, sync.json (per-device clock fits, IMU scales and time offsets). Every stream carries capture_ts_ns on one session clock.
Data collection logs logs/recording/ (the rig's take record, per-camera capture stats) and logs/processing/ (the reconstruction log and report, the frame check)

World frame: RDF. +x right (the left-to-right gripper direction, horizontal), +y down (gravity, from the head IMU), +z forward = x × y. The origin is the head's position at the first frame.

Downloads last month
205