The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: ArrowInvalid
Message: Column 1 named states expected length 732 but got length 600
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, 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/hdf5/hdf5.py", line 83, in _generate_tables
pa_table = _recursive_load_arrays(h5, self.info.features, start, end)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 279, in _recursive_load_arrays
return pa.Table.from_pydict(batch_dict)
~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^
File "pyarrow/table.pxi", line 1985, in pyarrow.lib._Tabular.from_pydict
File "pyarrow/table.pxi", line 6403, in pyarrow.lib._from_pydict
return cls.from_arrays(arrays, names, metadata=metadata)
File "pyarrow/table.pxi", line 4913, in pyarrow.lib.Table.from_arrays
result.validate()
File "pyarrow/table.pxi", line 4257, in pyarrow.lib.Table.validate
check_status(self.table.Validate())
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Column 1 named states expected length 732 but got length 600Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
SynTactiq — Synthetic Tactile Contact Dataset (Free Sample, 200K contacts)
Free sample of 608 MuJoCo simulation runs covering 200,176 contact samples across 13 materials (abs, anodized_aluminum, carbon_fiber, fabric, glass, industrial_rubber, plastic, rubber, silicone, silicone_ecoflex, silicone_pdms, steel, wood) and 5 contact scenarios (force_press, impact, shear, stick_slip, torsion). Each run is an HDF5 file with per-contact-point forces (normal + tangential), friction coefficients and contact geometry sampled at the simulator rate (240 Hz).
What is this dataset for?
- Pre-training tactile models (slip detection, force estimation, contact localization) before touching expensive real hardware such as Meta DIGIT or GelSight sensors.
- Sim-to-real research: the SynTactiq benchmark suite measures exactly how much real data synthetic pre-training saves on real DIGIT slip detection.
- Dataset curation / ML research: physically-consistent contact dynamics, documented HDF5 schema, checksums and split metadata out of the box.
What's inside
| Field | Description |
|---|---|
contacts/normal_force |
Normal force per contact point [N] |
contacts/tangential_force_1/2 |
Tangential shear components [N] |
contacts/friction_coefficients |
Static/dynamic friction per contact |
contacts/position |
Contact point position on the sensor surface |
Every package ships manifest.json (per-run provenance), SHA256SUMS (checksums), LICENSE (CC BY-NC 4.0) and split metadata.
Quick start
import h5py, glob
for path in sorted(glob.glob('*.h5'))[:3]:
with h5py.File(path, 'r') as f:
fn = f['contacts/normal_force'][:] # (T,) normal force
ft1 = f['contacts/tangential_force_1'][:] # (T,) shear x
mu = f['contacts/friction_coefficients'][:] # friction
print(path, fn.shape, 'peak Fn =', fn.max())
A slip label can be derived with the Coulomb friction-cone criterion (|Ft| >= mu * Fn), the same used in the public SynTactiq sim-to-real benchmark.
Intended uses & limitations
- Intended: research pre-training, domain-randomization studies, sim-to-real evaluation, teaching.
- Not intended: safety-critical control without real-sensor validation; the data is synthetic (MuJoCo), generated with physically-consistent parameters but not measured on hardware.
- License: CC BY-NC 4.0 — free for research and evaluation. Commercial training licenses are sold separately (see SynTactiq storefront / contact).
Citation
If this dataset helps your research, please cite it:
@dataset{sardella_2026_syntactiq_sample,
author = {Sardella, Jacopo},
title = {SynTactiq Tactile Contact Dataset -- Free Sample
(200K synthetic contacts, v0.2.0)},
year = 2026,
publisher = {Zenodo},
doi = {10.5281/zenodo.22277316},
url = {https://zenodo.org/records/22277316}
}
DOI: 10.5281/zenodo.22277316 · Code & sim-to-real benchmark: https://github.com/Jacopos311/SynTactiq
Package version: v0.2.0 · Runs: 608 · Contacts: 200,176 · License: CC BY-NC 4.0 (https://creativecommons.org/licenses/by-nc/4.0/)
- Downloads last month
- 268