Data Explorer

If you are interested in our dataset, please first read the readme.txt ! feel free to contact with us!

Human Hand: Wave Patterns and Perception This dataset is part of a larger repository (DOI: 10.5281/zenodo.5248082) which houses links to the data used in the publication "Shear Shock Waves Mediate Haptic Holography via Focused Ultrasound" ( Reardon et al., 2023 ). If you use these data please cite our publication ( http://www.science.org/doi/10.1126/sciadv.adf2037 ). This dataset contains the in vivo response of a single participant's hand to focused ultrasound (UHEV1, Ultrahaptics) scanned in a zigzag path from the wrist to the distal end of digit 2 (and vice-versa). The data is provided as…

Coding dynamic tactile information in spike timing is essential to human haptic exploration and dexterous object manipulation. Conventional electronic skins generate frames of tactile signals upon interaction with objects and are unfortunately ill-suited for efficient coding of temporal information and rapid feature extraction. Here, we report a neuromorphic tactile system that uses spike timing, especially the first-spike timing, to code dynamic tactile information about touch and grasp. This strategy enables the system to seamlessly code highly dynamic information with millisecond temporal…

This dataset comprises tactile data collected from three sensors: 36,630 probe images from a DIGIT sensor across 1,221 distinct probe locations; 36,270 probe images from a GelSight Mini sensor across 1,209 distinct probe locations; and 3,780 readings from a ReSkin magnetic tactile sensor across 189 distinct probe locations. The dataset also includes pre-trained model weights for 3D depth map reconstruction on our DIGIT and GelSight Mini sensors, as well as for contact location prediction on our ReSkin sensor.

This dataset contains the tactile data corresponding to 22 subjects, 12 women and 10 men, performing 21 different activities of daily living. The dataset was obtained using the Grip System, with sensor 4256E from Tekscan, Inc., USA. The tasks corresponded to common activities of daily living (ADL), selected to cover most of the areas related to hand function of the different chapters of the International Classification of Functioning, Disability and Health (ICF). This open dataset is, to our knowledge, the largest and most diverse database with tactile information of the human hand in ADL. Th…

Contains reading of single braille letters by sliding a robotic tactile sensor over each letter of the braille alphabet, including space. Braille letters are part of the dataset in the STL file format. Tactile data is pulled with 40Hz and saved in the pandas format and dumped in the pickle format. Sigma-delta modulation was used to create event-driven (spike-based) binary discrete-time data streams with four different encoding threshold. For more details please see the linked Github repository .

The pHRI-DIGI-TACT dataset provides time-series data captured from an uninstrumented, 3D-printed touchpad interface mounted on the end-effector of a collaborative robot. Users can draw digits (0-9) on the touchpad by sliding their finger, and the robot's joint torques are recorded. This dataset facilitates research in intuitive robot control, physical human-robot collaboration, tactile sensing, machine learning for gesture recognition, and digit recognition. It leverages the intrinsic tactile sensing capabilities of a KUKA LBR iiwa 14 R820 robot, interpreting touch events on the touchpad usin…

This dataset is the supplementary material of the IROS 2021 article "Organization and Understanding of a Tactile Information Dataset TacAct During Physical Human-Robot Interactions". The dataset was collected by a flexible supercapacitor tactile sensor installed in an imitated mechanical arm device. In a 32 × 32 grid, the sensor data can be sampled at 100 Hz (100 frames per second). A total of 12 touch actions, namely, pull, squeeze, push, hold, grasp, poke, static drag, strong hit, soft slide, scratch, soft tap, and sliding drag, were recorded. A single-action collected from a subject consis…

The dataset "Flight_Experiments_Data.zip" represents flight experiments information of the SwarmTouch project. During the experiment, involved participants were given a task to navigate a formation of drones through the obstacles maze relying only on tactile or visual feedback. For the experiment, users overcame two different unknown setups of obstacles, each setup firstly with tactile and then with vision feedback (two trials with tactile and two with visual (without glove) feedback in total). The dataset "Patterns_Recognition_Data.zip " represents tactile patterns recognition experimental r…

Dataset generated during test and experimentation for the paper Graphical visualization of contact forces and hand movements during in-hand manipulation.

This dataset provides visual and tactile recordings collected from a robotic grasping platform to support research on multi-modal perception, robot grasping, and sensor fusion. It integrates visual data with high-resolution tactile signals, enabling studies in perception-driven manipulation and multimodal learning. The dataset was collected to validate the proposed multimodal fusion method described in our paper. It is publicly available to promote reproducibility and facilitate further research in robotic perception and control. 1. Data Acquisition Platform The data collection platform consi…

Tactile dataset generated using a newly proposed, object-agnostic, protocol - the Dense Protocol- that allows collecting tactile data for training robust slip detection models during robotic object grasping. The proposed protocol steps are described in detail in [1], and a summary is provided below. The DENSE protocol The DENSE protocol is split into three main stages: generation of a valid set of grasp poses for each object; robotic object grasp execution and tactile data collection; and data labelling. In this work, the grasps generated by the proposed protocol are performed using the EZGri…

Existing robotic systems have a clear tension between generality and precision. Deployed solutions for robotic manipulation tend to fall into the paradigm of one robot solving a single task, lacking precise generalization, i.e., the ability to solve many tasks without compromising on precision. This paper explores solutions for precise and general pick-and-place. In precise pick-and-place, i.e. kitting, the robot transforms an unstructured arrangement of objects into an organized arrangement, which can facilitate further manipulation. We propose simPLE (simulation to Pick Localize and PLacE)…

This dataset includes all the experimental and FE results presented in the IEEE ICECS 2019 paper "Development of Fully Shielded Soft Inductive Tactile Sensors" (DOI: 10.1109/ICECS46596.2019.8964922). URL of IEEE Xplore: https://ieeexplore.ieee.org/abstract/document/8964922 List of data in this dataset: Fig-2-FE modeling-FS-SITS.xlsx Fig-4-Exp_characterization-FS-SITS.xlsx Fig-5-Exp_Demo-FS-SITS.xlsx All the data included in this dataset were collected by Dr. Hongbo Wang. Contact person: Dr. Hongbo Wang, ustcwhb@gmail.com

This dataset contains the measurement data for figures (graphs) published in journal article: Grid-type transparent conductive thin films of carbon nanotubes as capacitive touch sensors by Ronja Valasma, Eva Bozo, Olli Pitkänen, Topias Järvinen, Aron Dombovari, Melinda Mohl, Gabriela Simone Lorite, Janos Kiss, Zoltan Konya and Krisztian Kordas Published 11 May 2020 • © 2020 The Author(s). Published by IOP Publishing Ltd Nanotechnology, Volume 31, Number 30 Citation: Ronja Valasma et al 2020 Nanotechnology 31 305303 DOI 10.1088/1361-6528/ab8590

Recordings of a 12 taxel fingertip at 100Hz sliding over 3D printed Braille letters. Event encoding by sigma-delta modulation with four different thresholds: 1, 2, 5, and 10. Setup is similar to the dataset presented in [Braille publication] which is available online at [link to old zenodo repo], but with increased sampling rate and improved sensors. Further details can be found in the following publication: [publication outstanding]

The pHRI-BTN-TACT dataset provides time-series data captured from an uninstrumented, 3D-printed touchpad interface mounted on the end-effector of a collaborative robot featuring nine virtual buttons. This dataset facilitates research in intuitive robot control, physical human-robot collaboration, tactile sensing, and machine learning for touch recognition. It contains data collected using the intrinsic tactile sensing capabilities of a KUKA LBR iiwa 7 R800 robot, leveraging its joint torque sensors to interpret touch events on the touchpad. The dataset is presented in two versions: pHRI-BTN-T…

Automatic robotic-poking tactile dataset used in "Towards Artificial Palpation: Representation Learning of Touch on Soft Bodies" (Neurips 2025).

This is a official code for the benchmark of the paper "VTDexManip: A Dataset and Benchmark for Visual-tactile Pretraining and Dexterous Manipulation with Reinforcement Learning" (ICLR 2025)

Automatic robotic-palpation tactile dataset used in "More with LESS – Local Scene Representations for Tactile Imaging" (RSS 2026). We have included the "data-primitive" here, as described in the paper. See https://zoharri.github.io/LESS/ for more details.

This dataset has been generated during the SAHR (Skill Acquisition in Humans and Robots) project at Learning Algorithms and Systems Laboratory ( LASA ), EPFL, Switzerland. This project has recorded subjects' manipulation as videos in four batches in total. Each data file corresponds contains the following sensor datas recorded from one subject in each recording batch: FingerTPS sensor recording (finger pressure sensor) ATI Force/Torque sensor recording (base force/torque sensor) DigiTacts sensor recording (tool pressure sensor) This dataset has been originally published on GitHub on Mar 18, 2…

Multimodal Manipulation Learning Database The dataset consists of data recordings for object manipulation with audio-tactile sensory feedback for object handover. It captures the auditory and tactile signals of a Kuka IIWA robot with an Allegro hand holding a plastic container containing different materials. The robot manipulates the container with vertical shaking and rotation motions. The data consists of force/pressure measurements on the Allegro hand using a Tekscan tactile skin sensor, auditory signals from a microphone, and the joints data of the IIWA robot and the Allegro hand joints.…

This dataset contains all the data created while researching the topic of remote palpation. This contains the data used in the following papers: 'Comparative Analysis of Model-Based Predictive Shared Control for Delayed Operation in Object Reaching and Recognition Tasks With Tactile Sensing' 'Soft Morphing Interface for Tactile Feedback in Remote Palpation' 'Magneto-Active Elastomer Filter for Tactile Sensing Augmentation Through Online Adaptive Stiffening' 'Soft Control Interface for Highly Dexterous Unilateral Remote Palpation' 'Comparative Study of Hand-Tracking and Traditional Control Int…

Automatic robotic-palpation tactile dataset used in "More with LESS – Local Scene Representations for Tactile Imaging" (RSS 2026). We have included here the "data-poke", which is described in the paper (union of "data-v1" and "data-orientation"). See https://zoharri.github.io/LESS/ for more details.

Human skin perceives physical stimuli applied to the body and mitigates the risk of physical interaction through its soft and resilient mechanical properties. Social robots would benefit from whole-body robotic skin (or tactile sensors) resembling human skin in realizing a safe, intuitive, and contact-rich interaction with humans. However, existing soft tactile sensors show several drawbacks (complex structure, poor scalability, and fragility), which limit their application in whole-body robotic skin. Here, we introduce biomimetic robotic skin based on hydrogel-elastomer hybrids and tomograph…

A novel visual-tactile dataset for robotic manipulation

The datasets contain the data related to the experiments carried out for the tactile sensor characterization, related to the publication: A. Cirillo, M. Costanzo, G. Laudante, and S. Pirozzi, “Tactile Sensors for Parallel Grippers: Design and Characterization,” Sensors, vol. 21, no. 5, art. 1915, Mar. 2021. (DOI: 10.3390/s21051915)

Fast-adapting type 1 (FA-1) and slow-adapting type 1 (SA-1) first-order neurons in the human tactile system have distal axons that branch in the skin and form many transduction sites, yielding receptive fields with many highly sensitive zones or 'subfields'. We previously demonstrated that this arrangement allows FA-1 and SA-1 neurons to signal the geometric features of touched objects, specifically the orientation of raised edges scanned with the fingertips. Here we show that such signaling operates for fine edge orientation differences (5–20°) and is stable across a broad range of scanning…

Raw data recorded for the publication "Tactile Convolutional Networks for Online Slip and Rotation Detection" in rosbag format. Each .bag file contains recordings of two tactile sensors sampled with 1kHz as "sensor_msgs/Image" and a third channel for data labeling in the format "sr_robot_msgs/UBI0All". Detailed informations about the preprocessing and labeling are in the publication.

Automatic robotic-palpation tactile dataset used in "More with LESS – Local Scene Representations for Tactile Imaging" (RSS 2026). We have included here the "multiple" set measurements (phantoms with multiple inclusions), which is described in the paper. See https://zoharri.github.io/LESS/ for more details.

The datasets contain data for the normalization of tactile voltages, the training dataset for the diameter classifier, and data recorded during the experiment related to the publication: A. Cirillo, G. Laudante, and S. Pirozzi, “Tactile sensor data interpretation for estimation of wire features,” Electronics, vol. 10, no. 12, art. 1458, June 2021. (DOI: 10.3390/electronics10121458)

Tactile sensing in the human body is achieved via the skin. This has inspired the fabrication of synthetic skins with pressure sensors for potential applications in robotics, bio-medicine, and human-machine interfaces. Tactile sensors based on magnetic elements are promising as they provide high sensitivity and a wide dynamic range. However, current magnetic tactile sensors mostly detect pressures of solid objects and operate at relatively high forces about 100 mN. Here, we address these limitations by manufacturing soft, stretchable, and hair-like structures that are permanently magnetized t…

A novel dataset for surface defecf detection for aerospace metal parts. The dataset contains vision and tactile images with labels. Please refer to the paper for full description.

Objective: To evaluate the effect of an oral stimulation program in preterm on the performance in the first oral feeding, oral feeding skills and transition time from tube to total oral intake. Study Designer: Double-blind randomized clinical trial including very preterm newborns. Congenital malformations, intracranial hemorrhage grade III or IV, bronchopulmonary dysplasia, and necrotizing enterocolitis were excluded. Intervention group (GI) received an oral stimulation program of tactile extra-, peri-, and intraoral tactile manipulation once a day for 15 minutes, during a 10-day period. Cont…

Raw data and figures used for the article "Self-excited Contact Resonance Operation of a Tactile Piezoresistive Cantilever Microprobe with Diamond Tip", published in the proceedings of Sensor and Measurement Science International 2021; 2021-05-03 - 2021-05-06; digital.

Tactile perception dataset, comprising of the DIGIT sliding over YCB objects with ground-truth pose.

Visuo-tactile dataset with GelSight and depth camera for YCB objects.

Episodic memories comprise diverse attributes of experience distributed across neocortical areas. The hippocampus is integral to rapidly binding these diffuse representations, as they occur, to be later reinstated. However, the nature of the information exchanged during this hippocampal-cortical dialogue remains poorly understood. A recent study has shown that the secondary motor cortex carries two types of representations: place cell-like activity, which were impaired by hippocampal lesions, and responses tied to visuo-tactile cues, which became more pronounced following hippocampal lesions.…

A curated, learning-friendly list of tactile sensing research resources for robotic manipulation: papers, datasets, benchmarks, simulation platforms, and frameworks, covering the integration of touch with Vision-Language-Action (VLA) models and World Models for contact-rich manipulation

This repository contains links to the data used in the publication "Tactile Displays Driven by Projected Light". If you use these data, please cite our publication found here: (DOI: https://doi.org/10.1126/scirobotics.adv1383 ) Abstract From Preprint Manuscript Tactile displays that lend tangible form to digital content could transform computing interactions. However, achieving the resolution, speed, and dynamic range needed for perceptual fidelity remains challenging. We present a dynamic tactile display that directly converts projected light into visible and tactile patterns via a photomech…

The dataset includes pr ecisely made Computer-Aided Design (CAD) models in .step format, representing essential components such as metallic frame, silicone pad, plastic case, plastic grid, and tactile board with integrated proximity sensor. Additionally, the dataset provides a complete assembly of multi-modal sensor model in .f3z format. Furthermore, the dataset includes essential design files for the electronic infrastructure, including a precisely engineered circuit schematic design file in .sch format and a printed circuit board layout design file in .brd format. Additionally, the dataset…

Humans can quickly adjust their grip force to a change in friction at the object-skin interface during dexterous manipulation in a precision grip. To perform this adjustment, they rely on the feedback of the mechanoreceptive afferents innervating the fingertip skin. Because these tactile afferents encode information related to skin deformation, the nature of the feedback signaling a change in friction must somehow originate from a difference in the way the skin deforms when manipulating objects of different frictions. To better characterize the origin of the underlying sensory events, we aske…

Self-made Tri-Modal Dataset. This dataset provides visual, tactile, and textual recordings collected from a robotic grasping platform to support research on multi-modal perception, robot grasping, and task-guided manipulation. It integrates RGB-D visual data, high-resolution tactile force data, and natural language instructions, enabling studies in perception-driven manipulation, cross-modal learning, and robust object recognition under manufacturing uncertainties. The dataset was collected to validate the proposed tri-modal fusion method described in our paper. It is publicly available to pr…

The dataset contain the data acquired from the multi-sensorized fingers developed in T5.1 and integrated into grippers used in IntelliMan UC3 and UC4. The data contain tactile data, proximity data and endoscopic camera data for the evaluation of sensor performance with respect to IntelliMan use cases requirements.

The dataset consists of tactile signals acquired from two flexible PVDF sensor arrays (32 sensors total) integrated into soft sensing caps mounted on a TIAGo robotic gripper while grasping daily-life objects. Six objects (glass bottle, TV remote controller, plastic bottle, strawberry, rigid apple, and semi-soft apple) were each class consists of 50 trials, resulting in 300 trials. The dataset was segmented into short temporal windows and used to recognize objects at the early stage of grasping using machine-learning and deep-learning models, enabling real-time tactile-based object classificat…

The dataset includes raw signals recorded by an 8-unit PVDF piezoelectric tactile sensor array integrated into the thumb finger of the Hannes prosthetic hand. During object grasping, the sensors record the electrical response generated across three mechanical events: Press, Hold, Release.

A curated awesome list for dexterous robot manipulation, tactile sensing, dexterous hands, robot learning, datasets, benchmarks, and simulators.

A curated list of papers, datasets, simulators, hardware platforms, and open-source resources for dexterous robot hands and hand-object manipulation.

The dataset consists of tactile signals acquired from an 8-sensor PVDF (P(VDF-TrFE)) array embedded in a soft finger cap worn on the fingertip during touch interactions. Data were collected while subjects performed press, slide, and release actions on two textured surfaces (smooth and rough), with a 6-axis force/torque sensor used as ground truth for contact and slippage events. Two subjects were involved in this experiment, with each one collecting 150 trials from each texture type, and the dataset was used to develop and evaluate a threshold-based real-time algorithm for detecting contact o…

The dataset provides the data recorded during the experiments described in the paper “Costanzo, M.; De Simone, M.; Federico, S.; Natale, C. Non-Prehensile Manipulation Actions and Visual 6D Pose Estimation for Fruit Grasping Based on Tactile Sensing. Robotics 2023, 12, 92. https://doi.org/10.3390/robotics12040092”

This is a official code for pretraining models of the paper "VTDexManip: A Dataset and Benchmark for Visual-tactile Pretraining and Dexterous Manipulation with Reinforcement Learning" (ICLR 2025)

The dataset includes raw signals recorded by an 8-unit PVDF (P(VDF-TrFE)) piezoelectric tactile sensor array mounted on a biomimetic fingertip during controlled texture-exploration experiments.

Excel file containing the raw data of a paper published in Annals of Forest Science Journal intitled « Visuo-tactile and topographic characterizations of finished surface quality of oak and beech wood surfaces by French consumers and industrials: acceptability thresholds for raised grain ». Data concern topographical parameters (roughness and waviness parameters) measured on varnished surfaces (with one coat and two coats) of oak and beech woods. These data were correlated with the visuo-tactile perceptions of raised grain which occurs on these varnished surfaces (with one coat),…

The dataset represents the transfer function of the tactile sensing device, describing the relationship between input normal force (measured with a calibrated load cell as ground truth) and the output sensor voltage. The transfer function was obtained under sinusoidal excitation using an electromechanical shaker, and it covers a frequency range from 1 Hz to 1 kHz.

Tactile data collected in the wild, during autonomous robotic pick-and-place tasks.

The dataset has been collected from 20 naturalistic textures. A SynTouch Biotac sensor with 19 channels has been adopted to acquire tactile information under a controlled environment in terms of applied sliding velocity and indentation force. Each texture has been recorded 50 times with a sampling rate of 100 Hz and sliding duration of 8 seconds.

Data collection code and machine learning training dataset for paper "Robust and flexible sliding tactile sensor for surface pattern perception and recognition"

Dataset Details This dataset contains paired tactile and force data, intended for use in predicting 3-axis normal and shear forces applied to the sensor's elastomer. We used three different indenter shapes to collect force-labeled data: hemisphere, sharp, and flat. To measure force ground truths, we employed the ATI nano17 force/torque sensor. The protocol consisted of applying a random normal load (up to 3N) followed by a shear load, achieved by sliding the probe 2mm on the… See the full description on the dataset page: https://huggingface.co/datasets/facebook/gelsight-force-estimation.

A GelSight-Based Tactile Dataset for Surface Micro-Defect Detection

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "Unitree_G1_Inspire", "total_episodes": 85, "total_frames": 20075, "total_tasks": 1, "total_videos": 170, "total_chunks": 1, "chunks_size": 1000, "fps": 30, "splits": { "train": "0:85" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/eunjuri/baseball_depth_tactile.