Other
RCT
Tactile generalization
Robot-collected touch–vision–language data with 29,279 tactile frames from three DIGIT sensors pressing 122 industrial reference materials, paired with force, material photographs, descriptors and controlled held-out splits.
Access links
Primary places to inspect, download, read about, or reproduce the dataset.
Quick facts
- Vehicle type
- Other
- Environment
- Not reported
- Data origin
- Real-world
- Sensor count
- 3
- Ground truth
- Not reported
- Calibration
- Not reported
- Annotations
- Available
Sensor overview
High-level sensor availability before the detailed sensor records below.
Ground truth
Reference scope, method, rate, coverage, and provenance.
Not reported
Calibration and synchronization
Reported calibration level, reproducibility signals, and supporting notes.
- Level
- Not reported
- Processed parameters
- Not reported
- Raw calibration data
- Not reported
- Calibration targets
- Not reported
Calibration data
Calibration reproducibility details are not reported.
Notes
Not reportedKnown or intentional limitations
Reported constraints and characteristics to check before using the dataset.
Controlled reference-material presses do not cover arbitrary object geometries, curved surfaces or dynamic exploratory motions.
One photograph and descriptor set supervise each material; there are no pose-conditioned visual targets or frame-specific tactile captions.
Contact-position coverage is imbalanced; position 5 covers 75 of 122 materials and later positions are progressively sparse.
Material images have multiple valid tactile positives; strict sample-level diagonal retrieval is inappropriate.
A material-ID mismatch limited historical diverse-selection training to 121 of 122 materials; the paper states released metadata are corrected and held-out test coverage is unaffected.
Cameras
digit-tactile
- Model
- DIGIT
- Setup
- Not Reported
- Count
- 3
- Effective cameras
- 3
- Modality
- Other
- Resolution
- Not reported
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Optical tactile frames from three DIGIT instances used in turn by a rotating robot-arm adapter. Each press typically gives 15–17 frames at 0.10 mm indentation steps. The 2048 × 1536 material photographs are separate material-level supervision; their resolution does not describe DIGIT frames.
IMUs
Not reported
GNSS
Not reported
LiDAR
Not reported
Radar
Not reported
Pitot tube
Not reported
Additional sensors
Other sensor
- Name
- Contact-force measurement
- Count
- 1
- Notes
- Force signals are released and contact force is recorded for each tactile frame; apparatus identity, number of sensing elements and time-based sampling rate are not established.
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- Not reported
- Class count
- Not reported
Semantic segmentation
Not reported
Instance segmentation
Not reported
Object detection
Not reported
Optical flow
Not reported
Depth ground truth
Not reported
Citation
Citation key and BibTeX kept at the end of the page for reference.
@misc{he2026rct,
title={RCT: A Robot-Collected Touch–Vision–Language Dataset for Tactile Generalization},
author={He, Jingbo and Färber, Michael and Calandra, Roberto},
year={2026},
eprint={2606.31694},
archivePrefix={arXiv},
primaryClass={cs.RO},
url={https://arxiv.org/abs/2606.31694v1}
}
Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy