{
  "version": "v2026.11.17",
  "compiledAt": "2026-11-17",
  "scope": "Current RoboSkin tactile directory only; not an industry census. This edition corrects the FreeTacMan code-license description: the official README and root LICENSE identify Apache-2.0, while the provider dataset card labels data MIT. Public documentation only; no dataset payload download, code execution or rights clearance. Other records retain their existing access and scientific-source review dates.",
  "snapshot": "/releases/datasets/v2026.11.17.json",
  "count": 27,
  "candidates": [
    {
      "id": "star",
      "name": "STAR paper-associated dexterous manipulation data",
      "institution": [
        "Shanghai Innovation Institute",
        "Agibot"
      ],
      "year": 2026,
      "authors": [
        "Xiangcheng Liu",
        "Tianhao Wu",
        "Le Zheng",
        "Yidong Wang",
        "Bowen Jiang",
        "Mingjie Pan",
        "Xinlin Ren",
        "Yi Liu",
        "Jianlan Luo"
      ],
      "firstPublished": "2026-09-11",
      "paperVersion": "arXiv v1 (2609.12549v1)",
      "dataOrigin": "real",
      "robot": [
        "Bimanual wheeled mobile robot; two hands with 10 actuated DoF and 16 total DoF each; commercial model name unknown"
      ],
      "sensor": [
        "268-dimensional piezoresistive tactile sensing per hand",
        "Two wrist RGB cameras and one head RGB camera"
      ],
      "modalities": [
        "Three-view RGB",
        "Bilateral tactile signals",
        "Robot state",
        "Command actions",
        "Task-level language instructions"
      ],
      "sampleCount": "Author-reported 200 hours, 10,576 trajectories and 65 tasks. No hosted-file count verified.",
      "tasks": [
        "Dexterous multi-finger manipulation",
        "Pick-and-place and insertion",
        "Long-horizon manipulation"
      ],
      "objectCategories": "Task-specific household and manipulation objects; exhaustive inventory and unique object count unknown.",
      "dataFormat": "Paper describes timestamp alignment to 30 Hz; this is not the raw tactile sampling rate. Command actions use end-effector poses and hand joints (34 active dimensions, padded to 64 for the policy). Public package, file schema and train/validation/test manifests unknown.",
      "license": "Unknown: no dataset-file, code, weight or asset license verified. The arXiv article license does not license the described dataset.",
      "codeLicense": "Unknown",
      "assetLicense": "Unknown",
      "modelLicense": "Unknown",
      "generalization": "Evaluation objects are held out from 200-hour pretraining but used in 100 task-specific post-training trajectories per task; 20 evaluation trials per task use held-out initial configurations. Not unseen-object zero-shot generalization.",
      "paperUrl": "https://arxiv.org/abs/2609.12549v1",
      "projectUrl": "https://stardex-web.github.io/Star/",
      "researchUrl": "/guides/tactile-feedback-for-physical-ai",
      "availability": "Paper-linked data candidate only. Official project reopened on 2026-09-18; no dataset download, code release or dataset reuse license verified. Excluded from the dataset catalog count, available-file filters and Dataset structured data.",
      "sourceReviewed": "2026-09-18",
      "directoryRecordUrl": "https://roboskin.ai/datasets#candidate-star"
    }
  ],
  "statistics": {
    "total": 27,
    "access": {
      "gated": 3,
      "public-files": 10,
      "linked-unverified": 8,
      "announced": 3,
      "unknown": 3
    },
    "licenseDocumented": 13,
    "licenseUnknown": 14,
    "splitFilesListed": 1,
    "splitProtocolDescribed": 3,
    "splitUnknown": 23,
    "manifestInspected": 12
  },
  "selectionGuidanceReviewed": "2026-10-15",
  "entries": [
    {
      "id": "touchscale",
      "name": "TouchScale human visual-tactile dataset",
      "institution": [
        "TouchScale authors and 2077AI Data Foundation"
      ],
      "year": 2026,
      "firstPublished": "2026-10-07",
      "paperVersion": "arXiv v3 (2610.10288), November 2; host API revision 74914cc2e11d0a7f37ddbe5d57db4fe4e14a0a7e; revision-specific README inaccessible",
      "dataOrigin": "real",
      "robot": [
        "Human wearable collection; separate downstream robot experiments"
      ],
      "sensor": [
        "Bimanual tactile gloves",
        "Head RGB-D camera",
        "Wrist RGB cameras"
      ],
      "modalities": [
        "Human tactile observations",
        "Egocentric RGB-D",
        "Wrist RGB",
        "Language instructions"
      ],
      "sampleCount": "Paper: 500-hour study corpus. Reviewed public card: 100 hours / 15,324 episodes. Neither count was independently recounted from payloads.",
      "tasks": [
        "Visual-tactile representation learning",
        "Human contact supervision for later robot adaptation"
      ],
      "objectCategories": "Everyday human manipulation; exact contents of gated payloads not inspected.",
      "dataFormat": "Card describes WebDataset shards and per-episode HDF5 plus media/metadata. No loader or payload schema was executed or verified. Human recordings do not contain robot action labels.",
      "license": "Dataset card: CC BY-NC 4.0; noncommercial conditions apply. Gated access requires acceptance of host conditions and is not unrestricted reuse permission.",
      "licenseUrl": "https://huggingface.co/datasets/2077AIDataFoundation/TouchScale",
      "codeLicense": "Separate code terms not verified in this review.",
      "assetLicense": "Check object-asset terms separately before reuse.",
      "modelLicense": "No model-weight terms verified.",
      "generalization": "Human contact supervision is not robot-action supervision. The paper limits robot evaluation to one platform and four tasks; adaptation to another embodiment remains a separate experiment.",
      "paperUrl": "https://arxiv.org/html/2610.10288v3",
      "projectUrl": "https://touch-scale.github.io/",
      "githubUrl": "https://github.com/phai-lab/touchscale",
      "datasetUrl": "https://huggingface.co/datasets/2077AIDataFoundation/TouchScale",
      "researchUrl": "/news/touchscale-dataset-release-and-robot-learning",
      "availability": "Checked 2026-11-08: public card and host metadata only. The card still describes 100 hours and an earlier full-release plan; that plan does not verify 500-hour availability. No conditions accepted, gated data downloaded, episode decoded or model trained.",
      "sourceReviewed": "2026-11-08",
      "evidence": {
        "access": "gated",
        "accessCheckedAt": "2026-11-08",
        "licenseStatus": "documented",
        "dataLicense": "Public card labels data CC BY-NC 4.0; noncommercial conditions and host access terms apply. No conditions accepted in this review.",
        "dataLicenseUrl": "https://huggingface.co/datasets/2077AIDataFoundation/TouchScale",
        "publicFileScale": "Public card describes 100 hours and 15,324 episodes, distinct from the paper’s 500-hour study corpus; no payload recount.",
        "verificationScope": "Public page and API metadata inspected; host revision 74914cc2e11d0a7f37ddbe5d57db4fe4e14a0a7e. Revision-specific raw README returned HTTP 401. No authenticated files, episode decoding or training.",
        "splitStatus": "unknown",
        "splitDetails": "No released split manifest inspected.",
        "formatStatus": "described",
        "formatDetails": "Public card describes WebDataset shards and per-episode HDF5; no payload decoding.",
        "sources": [
          "https://huggingface.co/datasets/2077AIDataFoundation/TouchScale",
          "https://arxiv.org/html/2610.10288v3"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Visual-tactile representation learning; Human contact supervision for later robot adaptation. Collection: Human wearable collection; separate downstream robot experiments.",
        "useLimitations": "Human contact supervision is not robot-action supervision. The paper limits robot evaluation to one platform and four tasks; adaptation to another embodiment remains a separate experiment.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-touchscale"
    },
    {
      "id": "dexsuite-lift-scripted",
      "name": "DexSuite scripted Lift tactile subset",
      "institution": [
        "DexSuite authors, led by Mila and McGill University collaborators"
      ],
      "year": 2026,
      "paperVersion": "arXiv v1 (2610.37799); subset revision f4be6e95af40843da39cd5a88d8bd736393dd59c",
      "dataOrigin": "simulation",
      "robot": [
        "Simulated Franka arm with UMI-style tactile parallel gripper"
      ],
      "sensor": [
        "Simulated FlexiTac-style fingertip pads"
      ],
      "modalities": [
        "Simulated tactile force grids",
        "Front and wrist RGB video",
        "Robot and object state",
        "Controller actions"
      ],
      "sampleCount": "Metadata declares 200 episodes and 15,030 frames for this Lift subset. The paper-wide 4,700 episodes are a different scope; payload rows were not independently recounted.",
      "tasks": [
        "Scripted simulated lifting",
        "Contact-conditioned imitation learning research"
      ],
      "objectCategories": "Lift-task objects with randomized initialization, according to the subset card.",
      "dataFormat": "Inspected meta/info.json: LeRobot v3.0, 20 Hz, Parquet and MP4; left/right float32 force grids [12,32]; front/wrist RGB [224,224,3]. Action layout: 9 values, split into 7 manipulator pose-controller values and 2 gripper joint targets. Do not treat this as 9 joint angles.",
      "license": "This pinned dataset card labels the subset Apache-2.0. This is not a blanket license verification for the full corpus or simulator assets.",
      "licenseUrl": "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/blob/f4be6e95af40843da39cd5a88d8bd736393dd59c/README.md",
      "codeLicense": "Code license not verified; the repository route linked by PyPI returned 404 in this review.",
      "assetLicense": "Simulator and third-party asset terms require separate review.",
      "generalization": "Simulated contact is not physical sensor measurement. The card records pre-action observations and omits the final post-action state. Define terminal handling and a held-out episode split before training; the declared train split covers all 200 episodes.",
      "paperUrl": "https://arxiv.org/html/2610.37799v1",
      "projectUrl": "https://dexsuiteorg.github.io/",
      "datasetUrl": "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/tree/f4be6e95af40843da39cd5a88d8bd736393dd59c",
      "researchUrl": "/news/dexsuite-tactile-simulation-lerobot-data",
      "availability": "Checked 2026-10-31: public manifest, card, info.json, action_layout.json and collection_qa.json fetched at the pinned revision. Metadata inspected only; no Parquet/video payload download, loader execution, trajectory replay, training or hardware test.",
      "sourceReviewed": "2026-10-31",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-10-31",
        "licenseStatus": "documented",
        "dataLicense": "The pinned subset card labels the data Apache-2.0. Simulator software and third-party asset rights are separate and were not verified.",
        "dataLicenseUrl": "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/blob/f4be6e95af40843da39cd5a88d8bd736393dd59c/README.md",
        "publicFileScale": "21 listed paths. Metadata declares 200 episodes and 15,030 frames; payload rows were not independently recounted.",
        "verificationScope": "Pinned public file manifest, dataset card and JSON metadata inspected. No Parquet/video payload download, loader run, trajectory replay, training or hardware validation.",
        "splitStatus": "unknown",
        "splitDetails": "The inspected metadata declares train 0:200. No held-out evaluation partition or leakage-free split was validated.",
        "formatStatus": "described",
        "formatDetails": "JSON schema inspected: LeRobot v3.0, 20 Hz, [12,32] float32 left/right force grids and [224,224,3] RGB videos. Action layout has 7 manipulator pose-controller values plus 2 gripper joint targets. Payload decoding is untested.",
        "revision": "f4be6e95af40843da39cd5a88d8bd736393dd59c",
        "listedFileCount": 21,
        "sources": [
          "https://arxiv.org/html/2610.37799v1",
          "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/tree/f4be6e95af40843da39cd5a88d8bd736393dd59c",
          "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/blob/f4be6e95af40843da39cd5a88d8bd736393dd59c/meta/info.json",
          "https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted/blob/f4be6e95af40843da39cd5a88d8bd736393dd59c/meta/action_layout.json"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Scripted simulated lifting; Contact-conditioned imitation learning research. Collection: Simulated Franka arm with UMI-style tactile parallel gripper.",
        "useLimitations": "Simulated contact is not physical sensor measurement. The card records pre-action observations and omits the final post-action state. Define terminal handling and a held-out episode split before training; the declared train split covers all 200 episodes.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-dexsuite-lift-scripted"
    },
    {
      "institution": [
        "Cornell University EmPRISE Lab and paper collaborators"
      ],
      "year": 2025,
      "dataOrigin": "real",
      "sourceReviewed": "2026-10-27",
      "modelLicense": "Model-weight terms not verified in this documentation review.",
      "id": "clamp",
      "name": "CLAMP multimodal haptic dataset",
      "paperVersion": "CoRL 2025; arXiv v1 (2505.21495)",
      "robot": [
        "Human-operated sensorized reacher-grabber"
      ],
      "sensor": [
        "Active and passive thermal sensors",
        "Normal-force sensing resistor",
        "Contact microphone for vibration",
        "Two IMUs"
      ],
      "modalities": [
        "Thermal",
        "Normal-force signal",
        "Vibration",
        "Proprioception",
        "Object images",
        "Spoken annotations and derived language labels"
      ],
      "sampleCount": "Authors report 12.3 million samples across 5,357 household objects from 16 devices and 41 participants. These are paper totals, not independently recounted released files.",
      "tasks": [
        "Material recognition",
        "Compliance recognition",
        "Multimodal object perception"
      ],
      "objectCategories": "Household objects in the authors' collection.",
      "dataFormat": "The repository describes raw data and a filtered NPZ with precomputed visual predictions. The paper uses processed sensor features; its force input is an uncalibrated signal, not a verified Newton-valued release. File schema and payload completeness remain unverified.",
      "license": "Dataset license unverified: the official Dataverse route could not be inspected. The code license does not grant data reuse rights.",
      "codeLicense": "Repository code is labeled BSD-3-Clause; dataset and third-party terms must be checked separately.",
      "generalization": "Human collection and author-reported robot fine-tuning are different evidence. Sensor-specific preprocessing and embodiment adaptation are required; RoboSkin has not reproduced perception or manipulation results.",
      "paperUrl": "https://arxiv.org/html/2505.21495v1",
      "projectUrl": "https://emprise.cs.cornell.edu/clamp/",
      "githubUrl": "https://github.com/empriselab/CLAMP",
      "datasetUrl": "https://doi.org/10.7910/DVN/HNS2Z4",
      "availability": "Checked 2026-10-27: official project and repository link a Dataverse dataset. The data landing route was inaccessible in this review; no verification challenge was solved. File manifest, dataset license and payload remain unverified. No download, installation or experiment performed.",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-10-27",
        "licenseStatus": "unknown",
        "dataLicense": "Dataset license unverified: the official Dataverse route could not be inspected. The code license does not grant data reuse rights.",
        "publicFileScale": "No object manifest or payload independently inspected; counts are source-reported.",
        "verificationScope": "Official documentation review only; no access request, agreement acceptance, dataset download or reproduction.",
        "splitStatus": "unknown",
        "splitDetails": "Released file-level splits and leakage not verified.",
        "formatStatus": "described",
        "formatDetails": "The repository describes raw data and a filtered NPZ with precomputed visual predictions. The paper uses processed sensor features; its force input is an uncalibrated signal, not a verified Newton-valued release. File schema and payload completeness remain unverified.",
        "sources": [
          "https://arxiv.org/html/2505.21495v1",
          "https://emprise.cs.cornell.edu/clamp/",
          "https://doi.org/10.7910/DVN/HNS2Z4",
          "https://github.com/empriselab/CLAMP"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Material recognition; Compliance recognition; Multimodal object perception. Collection: Human-operated sensorized reacher-grabber.",
        "useLimitations": "Human collection and author-reported robot fine-tuning are different evidence. Sensor-specific preprocessing and embodiment adaptation are required; RoboSkin has not reproduced perception or manipulation results.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-clamp"
    },
    {
      "institution": [
        "Cornell University EmPRISE Lab and paper collaborators"
      ],
      "year": 2025,
      "dataOrigin": "real",
      "sourceReviewed": "2026-10-27",
      "modelLicense": "Model-weight terms not verified in this documentation review.",
      "id": "openrobocare",
      "name": "OpenRoboCare expert caregiving demonstrations",
      "paperVersion": "IROS 2025",
      "robot": [
        "Occupational therapists demonstrating on two manikins; not recorded robot trajectories"
      ],
      "sensor": [
        "Manikin pressure skin: 44 readings per frame",
        "RGB-D cameras",
        "Gaze and pose tracking"
      ],
      "modalities": [
        "Pressure JSON",
        "RGB-D",
        "Eye gaze",
        "2D/3D pose",
        "Hierarchical action annotations",
        "First/third-person video"
      ],
      "sampleCount": "Official sources report 21 occupational therapists, 15 tasks, 315 sessions and 19.8 hours. Counts are provider-reported, not independently audited.",
      "tasks": [
        "Caregiving activity perception",
        "Human demonstration analysis",
        "Multimodal contact and pose research"
      ],
      "objectCategories": "Caregiving activities performed on manikins, not clinical trials with patients.",
      "dataFormat": "Documentation describes pressure JSON, RGB JPEG, depth PNG, gaze/action CSV and pose JSON. Not every session contains every modality; inspect timestamps and missing files before alignment. Some 3D poses are reconstructed from 2D annotations and may contain noise.",
      "license": "Documentation names BSD-3-Clause but additionally limits use to research and education, prohibits re-identification, requires citation and participant privacy, and disallows redistribution without permission. AWS describes academic/non-commercial use. Do not treat this as unrestricted commercial permission.",
      "licenseUrl": "https://emprise.cs.cornell.edu/robo-care/docs/download.html",
      "codeLicense": "Separate software licensing not verified; the dataset documentation is not a blanket code license.",
      "generalization": "Manikin demonstrations are not clinical efficacy or safety evidence, and human activity annotations are not robot action supervision. Robot embodiment mapping and patient-context validation remain separate work; RoboSkin has not run these data.",
      "paperUrl": "https://emprise.cs.cornell.edu/robo-care/",
      "projectUrl": "https://emprise.cs.cornell.edu/robo-care/docs/organization.html",
      "datasetUrl": "https://registry.opendata.aws/open-robo-care/",
      "availability": "Checked 2026-10-27: official docs and AWS registry document public unsigned S3 access. This review inspected documentation only, not a live object manifest or payload; no data downloaded and no terms accepted.",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-10-27",
        "licenseStatus": "documented",
        "dataLicense": "Documentation names BSD-3-Clause but additionally limits use to research and education, prohibits re-identification, requires citation and participant privacy, and disallows redistribution without permission. AWS describes academic/non-commercial use. Do not treat this as unrestricted commercial permission.",
        "publicFileScale": "No object manifest or payload independently inspected; counts are source-reported.",
        "verificationScope": "Official documentation review only; no access request, agreement acceptance, dataset download or reproduction.",
        "splitStatus": "unknown",
        "splitDetails": "Released file-level splits and leakage not verified.",
        "formatStatus": "described",
        "formatDetails": "Documentation describes pressure JSON, RGB JPEG, depth PNG, gaze/action CSV and pose JSON. Not every session contains every modality; inspect timestamps and missing files before alignment. Some 3D poses are reconstructed from 2D annotations and may contain noise.",
        "sources": [
          "https://emprise.cs.cornell.edu/robo-care/",
          "https://emprise.cs.cornell.edu/robo-care/docs/organization.html",
          "https://registry.opendata.aws/open-robo-care/",
          "https://emprise.cs.cornell.edu/robo-care/docs/download.html"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Caregiving activity perception; Human demonstration analysis; Multimodal contact and pose research. Collection: Occupational therapists demonstrating on two manikins; not recorded robot trajectories.",
        "useLimitations": "Manikin demonstrations are not clinical efficacy or safety evidence, and human activity annotations are not robot action supervision. Robot embodiment mapping and patient-context validation remain separate work; RoboSkin has not run these data.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-openrobocare"
    },
    {
      "id": "egoexplorer",
      "name": "EgoExplorer human vision-motion-touch mini release",
      "institution": [
        "InternRobotics and EgoExplorer paper collaborators"
      ],
      "year": 2026,
      "firstPublished": "2026-10-13",
      "paperVersion": "arXiv v1 (2610.18872)",
      "dataOrigin": "real",
      "robot": [
        "Human wearable collection rig; no verified robot-action training release"
      ],
      "sensor": [
        "Wuji tactile arrays",
        "Egocentric camera",
        "Body trackers and hand tracking"
      ],
      "modalities": [
        "Egocentric video",
        "Tracker poses",
        "Hand skeletons",
        "Wuji tactile observations",
        "Reconstructed SMPL-X",
        "English step captions"
      ],
      "sampleCount": "The paper reports a 100-hour corpus. The current card describes an approval-gated mini release of 18 episodes / 66 minutes; these are not the full paper corpus or independently recounted payloads.",
      "tasks": [
        "Human loco-manipulation observation",
        "Multimodal representation and retargeting research"
      ],
      "objectCategories": "Ten task templates across laboratory, household and industrial activities, according to the card.",
      "dataFormat": "The card describes LeRobot v3.0 at 30 Hz, video, five tracker poses, two 21-joint hand skeletons, 24 x 31 Wuji tactile arrays per hand and refined SMPL-X. Reconstructed body geometry is distinct from measured tactile signals. Archive alignment and completeness were not tested.",
      "license": "Dataset card: CC BY-NC 4.0. Noncommercial conditions apply; approval or download access is not commercial permission.",
      "licenseUrl": "https://creativecommons.org/licenses/by-nc/4.0/",
      "codeLicense": "Toolkit code license not verified; do not inherit the dataset license.",
      "modelLicense": "Model-weight and third-party body-model terms not verified.",
      "generalization": "Human observations require embodiment, action-space and sensor mapping before robot policy use. Retargeting demonstrations do not establish robot-action labels or validated transfer to another hand.",
      "paperUrl": "https://arxiv.org/abs/2610.18872",
      "githubUrl": "https://github.com/InternRobotics/EgoExplorer",
      "datasetUrl": "https://huggingface.co/datasets/InternRobotics/EgoExplorer",
      "availability": "Checked 2026-10-26: the public card describes approval-gated access to 18 episodes / 66 minutes and a separate full-collection waitlist. Toolkit commit 0de464fd576939c053356df9be8be1cabc41c8cc exposes capture and reconstruction code. No access request, agreement acceptance, payload download, toolkit execution or training was performed.",
      "sourceReviewed": "2026-10-26",
      "evidence": {
        "access": "gated",
        "accessCheckedAt": "2026-10-26",
        "licenseStatus": "documented",
        "dataLicense": "Dataset card: CC BY-NC 4.0. Noncommercial conditions apply; approval or download access is not commercial permission.",
        "dataLicenseUrl": "https://creativecommons.org/licenses/by-nc/4.0/",
        "publicFileScale": "Card-described mini release: 18 episodes / 66 minutes. Payload files gated and not inspected; distinct from the paper 100-hour corpus.",
        "verificationScope": "Public card, paper abstract and toolkit commit inspected. No gated access request, payload download, training or hardware run.",
        "splitStatus": "unknown",
        "splitDetails": "No file-level train/test partition or leakage audit verified.",
        "formatStatus": "described",
        "formatDetails": "The card describes LeRobot v3.0 at 30 Hz, video, five tracker poses, two 21-joint hand skeletons, 24 x 31 Wuji tactile arrays per hand and refined SMPL-X. Reconstructed body geometry is distinct from measured tactile signals. Archive alignment and completeness were not tested.",
        "sources": [
          "https://arxiv.org/abs/2610.18872",
          "https://huggingface.co/datasets/InternRobotics/EgoExplorer",
          "https://github.com/InternRobotics/EgoExplorer/commit/0de464fd576939c053356df9be8be1cabc41c8cc"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Human loco-manipulation observation; Multimodal representation and retargeting research. Collection: Human wearable collection rig; no verified robot-action training release.",
        "useLimitations": "Human observations require embodiment, action-space and sensor mapping before robot policy use. Retargeting demonstrations do not establish robot-action labels or validated transfer to another hand.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-egoexplorer"
    },
    {
      "id": "tacclaw-infinity",
      "name": "TacClaw-Infinity / Daimon-Infinity public release",
      "institution": [
        "Daimon Robotics and the TacClaw-Infinity paper collaborators"
      ],
      "year": 2026,
      "firstPublished": "2026-10-14",
      "paperVersion": "arXiv v1 (2610.20320)",
      "dataOrigin": "real",
      "robot": [
        "TacClaw handheld collection interface; public host also contains DM-DataClaw and DM-DataDex material"
      ],
      "sensor": [
        "Fingertip tactile sensors and visual observation channels described by the authors"
      ],
      "modalities": [
        "Vision",
        "Tactile observations",
        "Language annotations",
        "End-effector trajectories"
      ],
      "sampleCount": "The paper reports over 10,000 hours, 1 million episodes and 1 billion frames across more than 1,000 task types. The hosting card separately describes a first 1,000-hour release. Neither the paper totals nor displayed storage bytes are an independently verified count of downloadable, complete episodes.",
      "tasks": [
        "Contact-rich manipulation demonstrations",
        "Vision-tactile-language-action learning",
        "Cross-embodiment transfer research"
      ],
      "objectCategories": "Diverse manipulation tasks reported by the authors; a deduplicated object inventory for the inspected public files was not verified.",
      "dataFormat": "The card documents a LeRobot-style layout. Public folders include DM-DataClaw, DataDex_Datasets and TacExo. Payload completeness, alignment and file-level train/test membership were not validated.",
      "license": "The hosting card labels the dataset CC BY-NC-SA 4.0: attribution, noncommercial use and share-alike conditions apply. Public download visibility is not a commercial-use permission.",
      "licenseUrl": "https://creativecommons.org/licenses/by-nc-sa/4.0/",
      "codeLicense": "Not verified in this data-access review; dataset terms do not establish code terms.",
      "modelLicense": "No model-weight license verified in this review.",
      "generalization": "The authors report transfer across three heterogeneous robots. RoboSkin has not reproduced that result. Human-collected end-effector trajectories require task, action-space and sensor compatibility checks before use on a different robot.",
      "paperUrl": "https://arxiv.org/abs/2610.20320v1",
      "projectUrl": "https://daimon-robotics.github.io/tacclaw-infinity-page/",
      "githubUrl": "https://github.com/daimon-robotics/TacClaw-Infinity",
      "datasetUrl": "https://huggingface.co/datasets/daimonrobotics/Daimon-Infinity",
      "availability": "Checked 2026-10-16: public file listings are visible at revision 27b2420cf9f516b3456d01418e0ed01b74c8a2b5. The card describes the first 1,000 hours, while the paper reports a larger corpus. The project page still marks release links Coming soon, so public-host files do not establish full paper-corpus release. We inspected documentation and directory listings only; no bulk download, checksum audit, training or hardware experiment was performed.",
      "sourceReviewed": "2026-10-16",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-10-16",
        "licenseStatus": "documented",
        "dataLicense": "The hosting card labels the dataset CC BY-NC-SA 4.0: attribution, noncommercial use and share-alike conditions apply. Public download visibility is not a commercial-use permission.",
        "publicFileScale": "Public directories observed: DM-DataClaw, DataDex_Datasets and TacExo. First 1,000 hours are claimed in the card; complete unique episode count and equivalence to the paper corpus remain unverified.",
        "verificationScope": "Paper abstract, hosting card, root and subset directory listings inspected. No payload download, checksum verification, split audit, training or hardware run.",
        "splitStatus": "unknown",
        "splitDetails": "No file-level split manifest validated. Do not infer leakage-free splits from task folder names.",
        "formatStatus": "described",
        "formatDetails": "The card documents a LeRobot-style layout. Public folders include DM-DataClaw, DataDex_Datasets and TacExo. Payload completeness, alignment and file-level train/test membership were not validated.",
        "revision": "27b2420cf9f516b3456d01418e0ed01b74c8a2b5",
        "sources": [
          "https://arxiv.org/abs/2610.20320v1",
          "https://huggingface.co/datasets/daimonrobotics/Daimon-Infinity",
          "https://huggingface.co/datasets/daimonrobotics/Daimon-Infinity/tree/27b2420cf9f516b3456d01418e0ed01b74c8a2b5",
          "https://creativecommons.org/licenses/by-nc-sa/4.0/"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Contact-rich manipulation demonstrations; Vision-tactile-language-action learning; Cross-embodiment transfer research. Collection: TacClaw handheld collection interface; public host also contains DM-DataClaw and DM-DataDex material.",
        "useLimitations": "The authors report transfer across three heterogeneous robots. RoboSkin has not reproduced that result. Human-collected end-effector trajectories require task, action-space and sensor compatibility checks before use on a different robot.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-tacclaw-infinity"
    },
    {
      "id": "romi-2k",
      "name": "ROMI-2K object-interaction dataset",
      "institution": [
        "Gaoling School of Artificial Intelligence, Renmin University of China",
        "Beijing Key Laboratory of Research on Large Models and Intelligent Governance",
        "Beijing Academy of Artificial Intelligence",
        "Beijing Jiaotong University",
        "State Key Laboratory of Multimedia Information Processing, Peking University",
        "AresoX"
      ],
      "year": 2026,
      "authors": [
        "Ruoxuan Feng",
        "Yutong Chen",
        "Ruihua Song",
        "Huan Yang",
        "Zhongyuan Wang",
        "Guocai Yao",
        "Di Hu"
      ],
      "firstPublished": "2026-10-03",
      "paperVersion": "arXiv v2 (2610.06955v2)",
      "dataOrigin": "real",
      "robot": [
        "Pika Sense handheld collector",
        "xArm 6 with Robotiq 2F-140 gripper for tabletop collection"
      ],
      "sensor": [
        "Two GelSight Mini optical tactile sensors",
        "RGB cameras and microphone",
        "Six-axis force/torque sensor on the robotic tabletop setup only; none on the handheld collector"
      ],
      "modalities": [
        "RGB",
        "Optical tactile images",
        "Audio",
        "Force/torque in the robotic tabletop subset",
        "Initial scene depth in the tabletop setup",
        "Object-property and content annotations"
      ],
      "sampleCount": "Paper v2 reports 1,657 handheld object–content combinations and 500 tabletop scenes (400 training, 100 test). These are different counting units, not a verified total of unique trajectories or objects.",
      "tasks": [
        "Object-centric active perception",
        "Material and content inference",
        "Multisensory property reasoning"
      ],
      "objectCategories": "Everyday objects and object–content combinations. Most tabletop collection objects are reused from the handheld subset; the test set adds 56 unseen objects according to the paper.",
      "dataFormat": "The paper describes synchronized visual, tactile and audio observations for six interactions: lift, press, collide, shake, rotate and squeeze. Force/torque belongs to tabletop collection. The complete released archive schema and file-level splits are unverified.",
      "license": "Unknown: no dataset-specific reuse license was identified in the reviewed official card or repository. A model-subdirectory Apache-2.0 license does not establish data reuse rights.",
      "codeLicense": "Unknown: no license file was identified in the reviewed GitHub tree.",
      "modelLicense": "Apache-2.0 text exists in the hosted ROMA-Qwen2.5-Omni-7B model directory; it does not license ROMI-2K or establish terms for every component.",
      "generalization": "The paper separates handheld training interactions from tabletop evaluation, with six predefined primitives on one arm–gripper setup. ROMA uses mean gravity force as text, not the full recorded force/torque stream. Dataset coverage is not proof of general manipulation-policy transfer.",
      "paperUrl": "https://arxiv.org/abs/2610.06955v2",
      "projectUrl": "https://gewu-lab.github.io/ROMA/",
      "githubUrl": "https://github.com/GeWu-Lab/ROMA",
      "datasetUrl": "https://huggingface.co/datasets/GeWu-Lab/ROMA",
      "availability": "Checked 2026-10-12: the official dataset card marks ROMI-2K and ROMA Bench as Coming soon. Checkpoints and one recorded demo scene are visible, but a complete dataset release is not confirmed. A checked release item in the code README conflicts with its coming-soon text. No full dataset download, inference or hardware run was performed.",
      "sourceReviewed": "2026-10-12",
      "evidence": {
        "access": "announced",
        "accessCheckedAt": "2026-10-12",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: no dataset-specific terms identified. The Apache-2.0 text in a model subdirectory does not license the dataset.",
        "publicFileScale": "Complete ROMI-2K data files not confirmed. The reviewed Hugging Face manifest lists model files; the GitHub tree contains one recorded demo scene.",
        "verificationScope": "Paper v2, official card, Hugging Face manifest and complete GitHub tree inspected. Coming-soon text conflicts with a checked README release item. No complete payload download, inference, training or hardware run.",
        "splitStatus": "protocol-described",
        "splitDetails": "Paper reports 400 training and 100 test tabletop scenes, with 56 additional unseen test objects. Released file-level split membership and leakage are unverified.",
        "formatStatus": "described",
        "formatDetails": "Visual, optical tactile and audio interactions; force/torque only in robotic tabletop collection. Full archive schema and completeness unverified.",
        "revision": "3069e802f83de142ad034964c3cd48779cd1e387",
        "sources": [
          "https://huggingface.co/datasets/GeWu-Lab/ROMA/tree/3069e802f83de142ad034964c3cd48779cd1e387",
          "https://github.com/GeWu-Lab/ROMA/tree/b016abf1ab8ef782ffe37f19b8094f447965daf1",
          "https://arxiv.org/html/2610.06955v2"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Object-centric active perception; Material and content inference; Multisensory property reasoning. Collection: Pika Sense handheld collector; xArm 6 with Robotiq 2F-140 gripper for tabletop collection.",
        "useLimitations": "The paper separates handheld training interactions from tabletop evaluation, with six predefined primitives on one arm–gripper setup. ROMA uses mean gravity force as text, not the full recorded force/torque stream. Dataset coverage is not proof of general manipulation-policy transfer.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-romi-2k"
    },
    {
      "id": "bench2dex",
      "name": "Bench2Dex simulation demonstrations",
      "institution": [
        "Shanghai Jiao Tong University",
        "Fudan University",
        "The University of Hong Kong",
        "Inspire Robots",
        "Zhongguancun Academy",
        "COWARobot Co. Ltd",
        "Nanyang Technological University"
      ],
      "year": 2026,
      "authors": [
        "Zhenjie Yang",
        "Yideng Zhang",
        "Dongjie Zhang",
        "Chenyu Jiang",
        "Xianshuai Liu",
        "Yufeng Li",
        "Zuhao Ge",
        "Xingyu Jiao",
        "Zheng Zhang",
        "Kaiyu He",
        "He Wang",
        "Yuwen Zhong",
        "Yi Deng",
        "Muyun Jiang",
        "Xianliang Huang",
        "Haisheng Su",
        "Donghang Zhang",
        "Jian Zhang",
        "Xue Yang",
        "Hongyang Li",
        "Zuxuan Wu",
        "Yu-Gang Jiang",
        "Xiaosong Jia",
        "Junchi Yan"
      ],
      "firstPublished": "2026-09-14",
      "paperVersion": "arXiv v1 (2609.15726v1)",
      "dataOrigin": "simulation",
      "robot": [
        "12 simulated bimanual dexterous arm–hand embodiments; 26 task–embodiment settings"
      ],
      "sensor": [
        "Surface-aligned simulated contact geometry; not measured output from a physical tactile sensor"
      ],
      "modalities": [
        "Multi-view RGB",
        "Joint state",
        "Commanded actions",
        "Object state",
        "Surface-aligned tactile maps",
        "Depth, 2D/3D boxes and occupancy when recorded"
      ],
      "sampleCount": "About 1,300 human-teleoperated simulation demonstrations in the v1 paper. Public hosting separately lists 5,200 HDF5 file paths; file paths are not a verified count of unique demonstrations.",
      "tasks": [
        "Bimanual tool use",
        "Articulated-object interaction",
        "Multi-stage manipulation",
        "Controlled scene perturbations"
      ],
      "objectCategories": "Task-specific tools, containers, articulated objects and assembly items; a deduplicated released object count is unknown.",
      "dataFormat": "HDF5 episodes. The documented schema aligns frame_index, timestamp_ns and sim_step; action commands precede the next simulator step and observations follow it. Actions include names, control mode, validity and source. Supported fields are not guaranteed in every payload; README says depth is omitted in the current storage workflow.",
      "license": "Repository root MIT applies to code; no separate dataset-file license or collection-wide model/asset license was verified.",
      "codeLicense": "MIT at repository revision f96a8b2b4eb475483af66e9e03916b35bc43f1be.",
      "assetLicense": "Unknown collection-wide terms; upstream robot and object asset licenses require separate review.",
      "modelLicense": "Unknown: no top-level weight license or model card verified; base-model terms may differ.",
      "generalization": "Four scene-perturbation channels (None, Equi., Inv., Full). Multiple hand support is not zero-shot cross-hand transfer; no physical-sensor or sim-to-real validation claimed.",
      "paperUrl": "https://arxiv.org/abs/2609.15726v1",
      "projectUrl": "https://bench2dex.github.io/",
      "githubUrl": "https://github.com/Bench2Dex/Bench2Dex",
      "datasetUrl": "https://huggingface.co/datasets/Bench2Dex/teleopdata/tree/b195787c65046083e6a43776b67bdf1389dfa3eb",
      "researchUrl": "/research/bench2dex-visuo-tactile-bimanual-benchmark-2026",
      "availability": "Public, ungated provider manifest verified on 2026-09-18. Dataset payloads were not downloaded; completeness, checksums, schema validity and permission to reuse data remain unverified. Paper-scale demonstrations and hosted file counts are different units.",
      "sourceReviewed": "2026-09-18",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-18",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: no dataset card or dataset-file license found in the reviewed provider manifest. Code MIT is not a dataset license.",
        "publicFileScale": "5,201 listed paths: 5,200 HDF5 files plus .gitattributes. Unique demonstrations, frames and bytes were not independently recounted.",
        "verificationScope": "Official project, v1 paper, repository root MIT license and provider manifests inspected. No payload download, checksum or HDF5 validation, simulation run or independent reproduction.",
        "splitStatus": "protocol-described",
        "splitDetails": "Paper defines deterministic task-partitioned evaluation seeds and None / Equi. / Inv. / Full perturbation channels, with 50 rollouts per task/channel/policy. Released training/validation/test split membership and leakage remain unknown; a directory name is not a verified split.",
        "formatStatus": "described",
        "formatDetails": "HDF5 episodes. The documented schema aligns frame_index, timestamp_ns and sim_step; action commands precede the next simulator step and observations follow it. Actions include names, control mode, validity and source. Supported fields are not guaranteed in every payload; README says depth is omitted in the current storage workflow.",
        "revision": "b195787c65046083e6a43776b67bdf1389dfa3eb",
        "listedFileCount": 5201,
        "sources": [
          "https://huggingface.co/datasets/Bench2Dex/teleopdata/tree/b195787c65046083e6a43776b67bdf1389dfa3eb",
          "https://github.com/Bench2Dex/Bench2Dex/blob/f96a8b2b4eb475483af66e9e03916b35bc43f1be/LICENSE",
          "https://arxiv.org/html/2609.15726v1"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Bimanual tool use; Articulated-object interaction; Multi-stage manipulation; Controlled scene perturbations. Collection: 12 simulated bimanual dexterous arm–hand embodiments; 26 task–embodiment settings.",
        "useLimitations": "Simulation demonstrations support controlled perturbation studies. Multiple supported hands do not establish zero-shot cross-hand or physical-sensor transfer.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-bench2dex"
    },
    {
      "id": "univtac-encoder-pretraining-corpus",
      "name": "UniVTAC Encoder Pretraining Corpus",
      "institution": [
        "ScaleLab, Shanghai Jiao Tong University",
        "D-Robotics",
        "ViTai Robotics",
        "The University of Hong Kong",
        "Nanjing University",
        "Shenzhen University",
        "Wuhan University",
        "Fudan University",
        "Tsinghua University"
      ],
      "year": 2026,
      "robot": [
        "UniVTAC simulated gripper contact-generation setup; not a physical-robot trajectory corpus"
      ],
      "sensor": [
        "Simulation-synthesized marker-based visuo-tactile observations"
      ],
      "modalities": [
        "Raw tactile RGB image with markers",
        "Marker-free tactile RGB image",
        "Gelpad depth map",
        "Projected 2D marker coordinates",
        "Object pose as 3D translation plus quaternion"
      ],
      "sampleCount": "The paper reports 205,826 simulation samples across 14 geometric primitives. The September 19 provider inventory separately lists 638 contact-pretraining HDF5 files across 14 non-empty shape directories. File counts are not frame counts or proof of equivalence to the paper corpus; neither collection is the task benchmark, 400 paper policy-training trajectories or 450 physical demonstrations.",
      "tasks": [
        "Shape reconstruction",
        "Contact-deformation prediction",
        "Marker-position prediction",
        "Relative object-pose regression",
        "Tactile representation pretraining"
      ],
      "objectCategories": "Fourteen geometric primitives ranging from convex forms such as spheres and cones to non-convex forms such as stars and cross-shapes, mounted on a standardized prism base for simulated contact generation.",
      "dataFormat": "The paper defines I_marked, I_pure, depth maps, projected 2D marker coordinates and a 7D object pose. The public contact/<shape>/hdf5/ paths were inventoried at revision e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd; payload arrays, frame totals and splits were not validated.",
      "license": "The provider dataset card labels the repository MIT. This is separate from the code root Apache-2.0 license and does not resolve third-party asset or checkpoint terms.",
      "paperUrl": "https://arxiv.org/abs/2602.10093",
      "projectUrl": "https://univtac.github.io/",
      "githubUrl": "https://github.com/univtac/UniVTAC",
      "licenseUrl": "https://opensource.org/license/mit",
      "datasetUrl": "https://huggingface.co/datasets/byml/UniVTAC/tree/e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd/contact",
      "researchUrl": "/research/univtac-platform-encoder-benchmark-2026",
      "availability": "Public contact-file listing verified on September 19, 2026. No standalone download establishing exact equivalence to the 205,826-sample paper corpus was verified; HDF5 payloads were not downloaded.",
      "sourceReviewed": "2026-09-19",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-19",
        "licenseStatus": "documented",
        "dataLicense": "MIT in the provider dataset-card metadata; third-party assets and checkpoint terms require separate review.",
        "dataLicenseUrl": "https://opensource.org/license/mit",
        "revision": "e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd",
        "publicFileScale": "638 listed contact HDF5 files across 14 non-empty shape directories. Equivalence to the paper's 205,826 contact samples is unverified; files are not frames.",
        "verificationScope": "Pinned provider file listing and code documentation inspected. No payload download, array validation, frame recount, training or hardware run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Public contact/<shape>/hdf5/ paths; the paper describes marked and marker-free RGB, depth, projected markers and 7D pose. Released payload arrays and paper-corpus equivalence remain unverified.",
        "sources": [
          "https://huggingface.co/datasets/byml/UniVTAC/tree/e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd/contact"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Shape reconstruction; Contact-deformation prediction; Marker-position prediction; Relative object-pose regression; Tactile representation pretraining. Collection: UniVTAC simulated gripper contact-generation setup; not a physical-robot trajectory corpus.",
        "useLimitations": "Use for simulated contact representations. Contact samples are not policy trajectories; public file counts do not establish equivalence to the paper corpus.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-univtac-encoder-pretraining-corpus"
    },
    {
      "id": "univtac-benchmark-dataset",
      "name": "UniVTAC Benchmark Dataset",
      "institution": [
        "ScaleLab, Shanghai Jiao Tong University",
        "D-Robotics",
        "ViTai Robotics",
        "The University of Hong Kong",
        "Nanjing University",
        "Shenzhen University",
        "Wuhan University",
        "Fudan University",
        "Tsinghua University"
      ],
      "year": 2026,
      "robot": [
        "Simulated Franka Panda with a parallel-jaw gripper and bilateral simulated GelSight Mini sensors"
      ],
      "sensor": [
        "Bilateral simulated GelSight Mini sensors through UniVTAC and TacEx"
      ],
      "modalities": [
        "Head RGB image",
        "Wrist RGB image",
        "Robot end-effector and joint state",
        "Robot action and episode metadata",
        "Bilateral tactile RGB",
        "Bilateral marker-overlaid tactile RGB",
        "Bilateral tactile depth",
        "Bilateral marker coordinates",
        "Bilateral tactile-sensor pose"
      ],
      "sampleCount": "September 19 file inventory at revision e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd lists 800 HDF5 paths under isaac45 and a separate 800 under isaac51, 100 per task. The historical August revision 172331dbbce95bc04c3e59b22f32dc72ba5561ae contained 800 HDF5 episodes, approximately 125.43 GB, with metadata marking 763 successful and 37 non-success episodes. Historical size/outcome counts were not revalidated for the current versions. Task files are distinct from encoder samples, policy-training trajectories and physical demonstrations.",
      "tasks": [
        "Lift Bottle",
        "Pull-out Key",
        "Lift Can",
        "Put Bottle in Shelf",
        "Insert Hole",
        "Insert HDMI",
        "Insert Tube",
        "Grasp Classify"
      ],
      "objectCategories": "Task-specific bottles, cans, keys and locks, shelf-placement objects, peg-and-hole components, HDMI connectors and ports, tube fixtures, and tactile-classification shapes as defined by the eight benchmark tasks.",
      "dataFormat": "Versioned isaac45/<task>/hdf5/ and isaac51/<task>/hdf5/ directories with task metadata. Collection documentation describes visual observations, robot state and bilateral tactile RGB, marker, depth and pose fields. Current payload schemas were not validated. The August Dataset Viewer CastError is historical; current viewer status was not retested.",
      "license": "MIT as displayed by the current Hugging Face dataset card. The official GitHub repository root LICENSE is separately Apache-2.0; its README’s MIT statement conflicts with that root license and is not used as the repository-license authority.",
      "licenseUrl": "https://opensource.org/license/mit",
      "paperUrl": "https://arxiv.org/abs/2602.10093",
      "projectUrl": "https://univtac.github.io/",
      "githubUrl": "https://github.com/univtac/UniVTAC",
      "datasetUrl": "https://huggingface.co/datasets/byml/UniVTAC",
      "researchUrl": "/research/univtac-platform-encoder-benchmark-2026",
      "availability": "Public file listing verified on September 19, 2026. Match main/Isaac Sim 4.5 to isaac45, or isaac51/Isaac Sim 5.1 to isaac51. The official repository states task data are not cross-compatible and released policy checkpoints target 4.5. See the research guide for the downloadable 30-row inventory and payload-verification limits.",
      "sourceReviewed": "2026-09-19",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-19",
        "licenseStatus": "documented",
        "dataLicense": "mit in the provider dataset-card metadata at the pinned revision; individual asset terms were not inspected.",
        "publicFileScale": "2,344 paths in the entire provider repository, shared with the contact and checkpoint collections. Task-only inventory: 800 HDF5 paths in isaac45 and 800 in isaac51, 100 per task in each version. Bytes, frames and unique demonstrations were not recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Versioned isaac45/<task>/hdf5/ and isaac51/<task>/hdf5/ paths with task metadata. Documentation describes head/wrist observations, robot state and bilateral tactile RGB, marker, depth and pose. Version compatibility and payload validity were not executed; the prior Dataset Viewer error was not retested.",
        "sources": [
          "https://huggingface.co/datasets/byml/UniVTAC/tree/e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd"
        ],
        "revision": "e1aee7b0c95543b535e0146b2de3ee1bc6ddaabd",
        "listedFileCount": 2344,
        "dataLicenseUrl": "https://opensource.org/license/mit"
      },
      "selectionGuidance": {
        "intendedUse": "Lift Bottle; Pull-out Key; Lift Can; Put Bottle in Shelf; Insert Hole; Insert HDMI; Insert Tube; Grasp Classify. Collection: Simulated Franka Panda with a parallel-jaw gripper and bilateral simulated GelSight Mini sensors.",
        "useLimitations": "Use the matching simulator version and task package. Simulated GelSight channels do not establish performance on a physical GelSight sensor.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-univtac-benchmark-dataset"
    },
    {
      "id": "t-rex",
      "name": "T-Rex Tactile-Reactive Dexterous Manipulation Dataset",
      "institution": [
        "UC Berkeley",
        "NVIDIA",
        "Stanford University",
        "Panasonic",
        "Sapienza University of Rome",
        "ItalAI"
      ],
      "year": 2026,
      "robot": [
        "Fixed-base bimanual Dexmate Vega-1 with two 22-DoF Sharpa Wave dexterous hands"
      ],
      "sensor": [
        "ZED X Mini head camera",
        "Two ZED X One S wrist cameras",
        "Ten fingertip image-based tactile sensors, five per hand"
      ],
      "modalities": [
        "Head RGB video",
        "Bilateral wrist RGB video",
        "Robot state",
        "Current and target joint positions",
        "Raw fingertip grayscale tactile video",
        "Estimated tactile deformation maps",
        "Estimated 6D fingertip wrenches",
        "Frame and episode metadata"
      ],
      "sampleCount": "The paper and official repository report a complete 100-hour collection with 7,700-plus trajectories, 22 motor primitives, and more than 200 objects. The public Hugging Face card specifies 5,464 episodes, 5,473,459 frames at 30 FPS (approximately 50 hours), 5,370 language-annotated trajectories, 22 motor primitives, and 207 objects. The public subset must not be described as the complete 100-hour corpus.",
      "tasks": [
        "Bimanual tactile-reactive motor primitives",
        "Dexterous manipulation",
        "Force-sensitive contact",
        "Deformable-object interaction",
        "Vision-language-action midtraining"
      ],
      "objectCategories": "The current public dataset card reports 207 unique canonical objects and publishes composition metadata, but RoboSkin did not independently audit every object label or collapse them into an inferred category taxonomy.",
      "dataFormat": "LeRobotDataset v3.0 with head and bilateral wrist videos, robot state, current and target joint positions, ten fingertip tactile streams, deformation maps, 6D wrenches, and episode metadata. The official repository provides selective download, inspection, and 3D replay examples.",
      "license": "MIT according to the current Hugging Face dataset card. Repository code is also MIT, but upstream robot assets, dependencies, and any files outside the public package retain their own terms.",
      "licenseUrl": "https://opensource.org/license/mit",
      "paperUrl": "https://arxiv.org/abs/2606.17055",
      "projectUrl": "https://tactile-reactive-dexterous.github.io/",
      "githubUrl": "https://github.com/ZhuoyangLiu2005/T-Rex",
      "datasetUrl": "https://huggingface.co/datasets/zekaiwang/trex_dataset",
      "availability": "The 5,464-episode, approximately 50-hour LeRobot v3.0 subset is publicly browsable and downloadable on Hugging Face. The official repository provides a dataset quickstart and states that the full 100-hour pretraining and midtraining corpus is not part of the current release.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-13",
        "licenseStatus": "documented",
        "dataLicense": "mit in the provider dataset-card metadata at the pinned revision; individual asset terms were not inspected.",
        "publicFileScale": "7,903 paths in the provider file manifest, including metadata. Bytes, frames and episodes were not independently recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "LeRobotDataset v3.0 with head and bilateral wrist videos, robot state, current and target joint positions, ten fingertip tactile streams, deformation maps, 6D wrenches, and episode metadata. The official repository provides selective download, inspection, and 3D replay examples.",
        "sources": [
          "https://huggingface.co/datasets/zekaiwang/trex_dataset/tree/bf0eb24c4b8bdd95752b553f0fc50e46a22f1cc8"
        ],
        "revision": "bf0eb24c4b8bdd95752b553f0fc50e46a22f1cc8",
        "listedFileCount": 7903,
        "dataLicenseUrl": "https://opensource.org/license/mit"
      },
      "selectionGuidance": {
        "intendedUse": "Bimanual tactile-reactive motor primitives; Dexterous manipulation; Force-sensitive contact; Deformable-object interaction; Vision-language-action midtraining. Collection: Fixed-base bimanual Dexmate Vega-1 with two 22-DoF Sharpa Wave dexterous hands.",
        "useLimitations": "Use for the recorded bimanual embodiment and tactile streams. The public subset is distinct from the complete paper corpus; transfer to another hand requires evaluation.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-t-rex"
    },
    {
      "id": "egotouch",
      "name": "EgoTouch",
      "institution": [
        "Harbin Institute of Technology, Shenzhen",
        "Meituan Academy of Robotics",
        "Tsinghua Shenzhen International Graduate School, Tsinghua University"
      ],
      "year": 2026,
      "robot": [
        "Human-wearable bimanual collection; no robot platform is used in the EgoTouch capture pipeline"
      ],
      "sensor": [
        "Head-mounted wide-angle camera",
        "Dual wrist cameras",
        "Bimanual pressure-sensing gloves with dense palm pressure maps",
        "Bimanual 3D hand-pose capture"
      ],
      "modalities": [
        "Head-mounted egocentric RGB",
        "Dual wrist-mounted RGB",
        "Bimanual 3D hand pose with 42 joints",
        "Continuous bilateral tactile pressure maps",
        "Camera and wrist poses",
        "Frame timestamps"
      ],
      "sampleCount": "208 manipulation tasks across 1,891 episodes in indoor and outdoor environments. The official repository reports approximately 2 million frames and more than 1,000 objects; those repository statistics are author-reported and the repository still warns that the dataset upload may be incomplete.",
      "tasks": [
        "Bimanual hand-object manipulation",
        "Egocentric vision-to-touch prediction",
        "Contact prediction under visual occlusion",
        "Multi-view tactile estimation"
      ],
      "objectCategories": "More than 1,000 objects across home, office, outdoor, retail, and workbench scenes are reported by the official repository; a complete reviewed object taxonomy was not verified.",
      "dataFormat": "The official repository documents variable-length 30 FPS HDF5 episodes containing three 640 x 480 RGB views, bimanual hand-pose arrays, bilateral pressure grids, camera poses, masks when available, metadata, and timestamps. The Hugging Face repository also exposes scene folders and split.json.",
      "license": "A separate dataset-file license was not verified. The official GitHub repository is MIT licensed, but that project-code license is not generalized here to the hosted EgoTouch data files.",
      "paperUrl": "https://arxiv.org/abs/2605.13083",
      "githubUrl": "https://github.com/Jianyi2004/TouchAnything",
      "datasetUrl": "https://huggingface.co/datasets/zhouzhoujy/EgoTouch",
      "availability": "The official Hugging Face repository exposes scene folders and split metadata. The official GitHub repository says the upload is in progress and currently accessible files may be incomplete, so completeness and the exact version must be checked before reuse.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-13",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "22,897 paths in the provider file manifest, including metadata. Bytes, frames and episodes were not independently recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "files-listed",
        "splitDetails": "Split-named files listed: split.json. Contents and leakage were not validated.",
        "formatStatus": "described",
        "formatDetails": "The official repository documents variable-length 30 FPS HDF5 episodes containing three 640 x 480 RGB views, bimanual hand-pose arrays, bilateral pressure grids, camera poses, masks when available, metadata, and timestamps. The Hugging Face repository also exposes scene folders and split.json.",
        "sources": [
          "https://huggingface.co/datasets/zhouzhoujy/EgoTouch/tree/cfdbb0ac31cc2af4247943820aa250575e7e6637"
        ],
        "revision": "cfdbb0ac31cc2af4247943820aa250575e7e6637",
        "listedFileCount": 22897
      },
      "selectionGuidance": {
        "intendedUse": "Bimanual hand-object manipulation; Egocentric vision-to-touch prediction; Contact prediction under visual occlusion; Multi-view tactile estimation. Collection: Human-wearable bimanual collection; no robot platform is used in the EgoTouch capture pipeline.",
        "useLimitations": "Human glove pressure and hand pose can inform representation or retargeting research. They are not robot action labels or calibrated signals for a different sensor.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-egotouch"
    },
    {
      "id": "prism-industrial-skill",
      "name": "PRISM: Precision and contact-rich Real-world Industrial Skill dataset with Multimodal sensing",
      "institution": [
        "State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University",
        "Delta Intelligence",
        "PKU-Wuhan Institute for Artificial Intelligence",
        "Hubei Humanoid Robot Innovation Center Co., Ltd.",
        "China Academy of Information and Communications Technology"
      ],
      "year": 2026,
      "robot": [
        "Dual Franka Emika Panda tracker platform",
        "Bimanual Realman RM75-6F exoskeleton platform",
        "LEJU upper-body humanoid VR platform"
      ],
      "sensor": [
        "Multi-view RGB-D cameras",
        "6DoF force/torque sensing",
        "Unnamed visuotactile gripper and visuotactile dexterous hand on a subset of Franka episodes"
      ],
      "modalities": [
        "Multi-view RGB",
        "Depth",
        "Visuotactile image when available",
        "6DoF force/torque when available",
        "Robot joint angle and torque",
        "End-effector Cartesian pose",
        "Gripper state",
        "Calibration and timestamps",
        "Human control signal"
      ],
      "sampleCount": "5,000+ robot trajectories paired with 5,000 human demonstrations; 45+ hours; 25+ industrial tasks; approximately 27M images across visual and visuotactile streams. The paper does not disclose the number or share of tactile-equipped episodes.",
      "tasks": [
        "Electronic component plug and unplug",
        "Caliper packaging",
        "Conveyor sorting",
        "NIST Assembly Task Board operations",
        "Industrial installation and assembly",
        "Long-horizon compositional procedures"
      ],
      "objectCategories": "Industrial components, assembly hardware, packaging items, conveyor-sorted materials, and task-board objects; the source does not provide one complete released object inventory.",
      "dataFormat": "The paper describes timestamped episodes in a cross-platform schema and conversion to LeRobot v3.0 for its experiments. The current linked host did not expose a readable manifest or payload in this review; final file layout and decoding remain unverified.",
      "license": "Dataset-file license not verified. The linked ModelScope page did not expose readable terms in this check; paper and code licenses are not data permission.",
      "paperUrl": "https://arxiv.org/abs/2608.17962",
      "projectUrl": "https://tengbo-yu.github.io/PRISM/",
      "datasetUrl": "https://modelscope.cn/datasets/Turboyuuuu/PRISM",
      "datasetUrlReviewed": "2026-11-04",
      "githubUrl": "https://github.com/Tengbo-Yu/PRISM",
      "availability": "Access checked November 4, 2026: the official project links a ModelScope dataset route, superseding the August 22 disabled-button observation. The target page did not expose a readable manifest or dataset-file terms. Download route linked; files, payload completeness, reuse license and loader execution remain unverified. Scientific-source fields retain their original review date.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-11-04",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Official project data link checked. Linked ModelScope page did not expose readable file manifest or terms; no payload download, checksum, loader, training or hardware test.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "The paper describes timestamped episodes in a cross-platform schema and conversion to LeRobot v3.0 for its experiments. The current linked host did not expose a readable manifest or payload in this review; final file layout and decoding remain unverified.",
        "sources": [
          "https://tengbo-yu.github.io/PRISM/",
          "https://modelscope.cn/datasets/Turboyuuuu/PRISM"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Electronic component plug and unplug; Caliper packaging; Conveyor sorting; NIST Assembly Task Board operations; Industrial installation and assembly; Long-horizon compositional procedures. Collection: Dual Franka Emika Panda tracker platform; Bimanual Realman RM75-6F exoskeleton platform; LEJU upper-body humanoid VR platform.",
        "useLimitations": "Contact-rich industrial collection is task and hardware dependent. Check the released action schema, force conventions and access conditions before planning policy training.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-prism-industrial-skill"
    },
    {
      "id": "softvtbench",
      "name": "SoftVTBench",
      "institution": [
        "Tuojing Intelligence",
        "Tsinghua University",
        "King's College London",
        "Southeast University",
        "Stevens Institute of Technology",
        "Hong Kong University of Science and Technology (Guangzhou)",
        "University of Manchester",
        "Simple AI",
        "Imperial College London",
        "Carnegie Mellon University",
        "Zhejiang University",
        "Beihang University",
        "University of Hong Kong"
      ],
      "year": 2026,
      "robot": [
        "Simulated Franka arm with Panda parallel-jaw gripper in Isaac Sim and Isaac Lab"
      ],
      "sensor": [
        "Simulated bilateral GelSight Mini profiles rendered through TacEx, Taxim, and FOTS"
      ],
      "modalities": [
        "Multi-view RGB",
        "Dual-finger tactile RGB",
        "Tactile marker motion",
        "Proprioception",
        "Language",
        "Continuous and binary gripper actions",
        "Evaluator-only FEM state"
      ],
      "sampleCount": "Latest paper: 4,000 expert demonstrations across 40 tasks and more than 50 assets. Hugging Face dataset-card revision fd2793a documents four subsets with 10 tasks and 100 successful demonstrations each, or 4,000 demonstrations total. The GitHub README last changed 2026-07-22 still lists an earlier 1,628-demonstration, 33-asset state, so the first-party release documents are not synchronized.",
      "tasks": [
        "Deformable-object grasp and placement",
        "Rigid-twin matched control tasks",
        "Object-variation generalization",
        "Spatial-variation generalization",
        "Vision-only and visuo-tactile policy comparison"
      ],
      "objectCategories": "Ten volumetric deformable manipulated-object designs and matched rigid twins; the latest paper reports more than 50 assets across the complete benchmark inventory.",
      "dataFormat": "HDF5 trajectories with synchronized policy observations, robot actions, simulator state, and evaluator fields, plus third-person, wrist, and bilateral tactile MP4 videos documented by the current dataset card.",
      "license": "Apache-2.0 as shown on the current Hugging Face dataset card; upstream simulator, robot, tactile-runtime, and third-party asset terms may differ",
      "licenseUrl": "https://www.apache.org/licenses/LICENSE-2.0",
      "paperUrl": "https://arxiv.org/abs/2608.18701",
      "projectUrl": "https://softvtbench.github.io/",
      "githubUrl": "https://github.com/TuojingAI/SoftVTBench",
      "datasetUrl": "https://huggingface.co/datasets/Arthur12137/SoftVTBench",
      "availability": "The official repository links public code and the Hugging Face dataset mirror. The Hugging Face card reviewed on 2026-08-22 documents 4,000 hosted demonstrations, while the older GitHub README still lists 1,628 demonstrations and 33 assets. Pin the exact dataset revision because the first-party release documents currently disagree.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-13",
        "licenseStatus": "documented",
        "dataLicense": "apache-2.0 in the provider dataset-card metadata at the pinned revision; individual asset terms were not inspected.",
        "publicFileScale": "16,060 paths in the provider file manifest, including metadata. Bytes, frames and episodes were not independently recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "HDF5 trajectories with synchronized policy observations, robot actions, simulator state, and evaluator fields, plus third-person, wrist, and bilateral tactile MP4 videos documented by the current dataset card.",
        "sources": [
          "https://huggingface.co/datasets/Arthur12137/SoftVTBench/tree/fd2793a19310b5ba4ac6518f9a17ff43d56f6651"
        ],
        "revision": "fd2793a19310b5ba4ac6518f9a17ff43d56f6651",
        "listedFileCount": 16060,
        "dataLicenseUrl": "https://www.apache.org/licenses/LICENSE-2.0"
      },
      "selectionGuidance": {
        "intendedUse": "Deformable-object grasp and placement; Rigid-twin matched control tasks; Object-variation generalization; Spatial-variation generalization; Vision-only and visuo-tactile policy comparison. Collection: Simulated Franka arm with Panda parallel-jaw gripper in Isaac Sim and Isaac Lab.",
        "useLimitations": "Soft-contact experiments depend on the reported objects, deformation and sensor setup. Do not extend reported success to arbitrary materials or contact forces.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-softvtbench"
    },
    {
      "id": "robotacdex",
      "name": "RoboTacDex: A Dexterous Visual-Tactile-Action Dataset for Humanoid Manipulation",
      "institution": [
        "Fudan University",
        "ByteDance Intelligent Creation",
        "Shanghai AI Laboratory",
        "The Chinese University of Hong Kong"
      ],
      "year": 2026,
      "robot": [
        "Unitree G1 humanoid with dual arms totaling 14 DoF and two BrainCo Revo2 Tactile dexterous hands totaling 12 hand DoF as counted by the paper; lower limbs and waist fixed during collection"
      ],
      "sensor": [
        "Four RGB-D cameras, including two Intel RealSense D435i cameras for the head and third-person views",
        "Two BrainCo Revo2 Tactile dexterous hands measuring fingertip normal force, tangential force and direction, and self-capacitance proximity"
      ],
      "modalities": [
        "Four-view RGB at 640 x 480",
        "Four-view depth at 640 x 480",
        "Arm and finger joint state",
        "Arm and finger joint action",
        "Bilateral fingertip normal force",
        "Bilateral fingertip tangential force and direction",
        "Self-capacitance proximity",
        "Natural-language semantic annotation"
      ],
      "sampleCount": "More than 6,000 physical-robot trajectories totaling approximately 25 hours, covering 19 tasks, an author-reported 23 skills, and 22 objects. Figure 4 exposes 22 discernible atomic-skill labels, so the 23-skill headline has an unresolved internal source discrepancy. Trajectories are recorded at 30 Hz; tactile and dexterous-hand joint-state messages are published at 100 Hz and recorded locally at 30 Hz with the other modalities.",
      "tasks": [
        "Basic grasping, placing, and pushing",
        "Articulated-object manipulation",
        "Dual-arm collaborative manipulation",
        "Fine manipulation",
        "Humanoid-interactive manipulation",
        "Pick and place a pear",
        "Turn a page",
        "Insert a book into a document bag",
        "Unscrew a bottle cap"
      ],
      "objectCategories": "Twenty-two objects spanning articulated, constrained, container, functional, and graspable categories, including kitchen food and office supplies as well as rigid and deformable objects. The paper does not publish a complete machine-readable object inventory.",
      "dataFormat": "The v1 paper describes 30 Hz synchronized trajectories containing four 640 x 480 RGB-D views, arm and finger states and actions, bilateral tactile signals, and semantic annotations. It does not state the public package format, directory schema, compression, checksums, or train-validation-test file layout.",
      "license": "Dataset-file license not stated. The arXiv article license does not license unreleased dataset files.",
      "paperUrl": "https://arxiv.org/abs/2606.31836",
      "availability": "The June 30, 2026 v1 paper says the dataset will be open-sourced soon. No official project page, repository, downloadable package, or dataset-file license was verified on 2026-08-22; record access as announced, not released.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "announced",
        "accessCheckedAt": "2026-08-22",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "The v1 paper describes 30 Hz synchronized trajectories containing four 640 x 480 RGB-D views, arm and finger states and actions, bilateral tactile signals, and semantic annotations. It does not state the public package format, directory schema, compression, checksums, or train-validation-test file layout.",
        "sources": [
          "https://arxiv.org/abs/2606.31836"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Basic grasping, placing, and pushing; Articulated-object manipulation; Dual-arm collaborative manipulation; Fine manipulation; Humanoid-interactive manipulation; Pick and place a pear; Turn a page; Insert a book into a document bag; Unscrew a bottle cap. Collection: Unitree G1 humanoid with dual arms totaling 14 DoF and two BrainCo Revo2 Tactile dexterous hands totaling 12 hand DoF as counted by the paper; lower limbs and waist fixed during collection.",
        "useLimitations": "A described collection is not a verified downloadable training package. Confirm release, action schema and dataset terms before committing to reproduction.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-robotacdex"
    },
    {
      "id": "ht-bench",
      "name": "HT-Bench",
      "institution": [
        "Beihang University",
        "Rimbot",
        "ShanghaiTech University",
        "Tsinghua University",
        "Chinese Academy of Sciences",
        "BUPT"
      ],
      "year": 2026,
      "robot": [
        "Single reported dexterous full-hand collection and downstream-evaluation platform; hardware model not stated in the reviewed HTML"
      ],
      "sensor": [
        "Single reported full-hand tactile sensing pipeline; hardware model not stated in the reviewed HTML"
      ],
      "modalities": [
        "Egocentric RGB",
        "Full-hand tactile pressure maps"
      ],
      "sampleCount": "10M RGB frames; 7.8M tactile frames; 226 tasks",
      "tasks": [
        "Tactile similarity retrieval",
        "Masked tactile inpainting",
        "Vision-to-tactile synthesis",
        "Tactile frame prediction"
      ],
      "objectCategories": "Home, electronics workbench, chemistry lab, retail, workbench, outdoor, and other scenes; object count not stated.",
      "dataFormat": "Synchronized egocentric RGB and full-hand tactile maps; the benchmark pipeline normalizes tactile maps to 224 x 224.",
      "license": "Dataset, model-weight, and code licenses not stated; the paper is CC BY 4.0, which does not establish a license for unreleased artifact files",
      "paperUrl": "https://arxiv.org/abs/2606.19161v2",
      "availability": "The v2 preprint says the authors will release all data, evaluation protocols, pretrained weights, and training/testing scripts. No dedicated downloadable dataset package, official repository, or artifact license was verified on 2026-08-22; record access as announced, not released.",
      "sourceReviewed": "2026-08-22",
      "evidence": {
        "access": "announced",
        "accessCheckedAt": "2026-08-22",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Synchronized egocentric RGB and full-hand tactile maps; the benchmark pipeline normalizes tactile maps to 224 x 224.",
        "sources": [
          "https://arxiv.org/abs/2606.19161v2"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Tactile similarity retrieval; Masked tactile inpainting; Vision-to-tactile synthesis; Tactile frame prediction. Collection: Single reported dexterous full-hand collection and downstream-evaluation platform; hardware model not stated in the reviewed HTML.",
        "useLimitations": "Full-hand prediction and retrieval tasks do not by themselves validate closed-loop control. Check whether data, splits and evaluation code are actually released.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-ht-bench"
    },
    {
      "id": "rct",
      "name": "RCT: Robotic Contact Tactile",
      "institution": [
        "TU Dresden",
        "ScaDS.AI Dresden/Leipzig",
        "LASR Lab"
      ],
      "year": 2026,
      "robot": [
        "Robot arm with rotating three-sensor adapter"
      ],
      "sensor": [
        "DIGIT vision-based tactile sensor"
      ],
      "modalities": [
        "Tactile image",
        "Material RGB image",
        "Language descriptors",
        "Normal force",
        "Indentation depth"
      ],
      "sampleCount": "29,279 tactile frames; 1,832 contact sequences; 122 materials; 3 DIGIT sensors",
      "tasks": [
        "Material generalization",
        "Touch-to-text retrieval",
        "Touch-to-vision retrieval",
        "Sensor-disjoint evaluation"
      ],
      "objectCategories": "122 industrial reference materials in 7 categories.",
      "dataFormat": "Ordered contact sequences with material, category, sensor, position, depth, force, image, and descriptor metadata.",
      "license": "CC BY 4.0 dataset; Apache-2.0 code",
      "licenseUrl": "https://creativecommons.org/licenses/by/4.0/",
      "paperUrl": "https://arxiv.org/abs/2606.31694",
      "projectUrl": "https://faerber-lab.github.io/RCT/",
      "githubUrl": "https://github.com/faerber-lab/RCT",
      "datasetUrl": "https://figshare.com/s/a5ed417ba6602ccad0f6",
      "availability": "Dataset, split tools, and evaluation code are publicly linked by the official project page.",
      "sourceReviewed": "2026-08-16",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-08-16",
        "licenseStatus": "documented",
        "dataLicense": "CC BY 4.0 dataset; Apache-2.0 applies to code, according to the existing project review.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "protocol-described",
        "splitDetails": "Existing source review documents split tools. File-level membership and train/test leakage were not independently validated.",
        "formatStatus": "described",
        "formatDetails": "Ordered contact sequences with material, category, sensor, position, depth, force, image, and descriptor metadata.",
        "sources": [
          "https://figshare.com/s/a5ed417ba6602ccad0f6"
        ],
        "dataLicenseUrl": "https://creativecommons.org/licenses/by/4.0/"
      },
      "selectionGuidance": {
        "intendedUse": "Material generalization; Touch-to-text retrieval; Touch-to-vision retrieval; Sensor-disjoint evaluation. Collection: Robot arm with rotating three-sensor adapter.",
        "useLimitations": "Use for material/contact representations and the documented sensor-disjoint protocol. Contact sequences do not supply a general manipulation-policy benchmark.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-rct"
    },
    {
      "id": "tactidex",
      "name": "TactiDex",
      "institution": [
        "ShanghaiTech University",
        "InstAdapt"
      ],
      "year": 2026,
      "robot": [
        "Human demonstration capture",
        "Bimanual Franka Inspire deployment"
      ],
      "sensor": [
        "Whole-hand tactile glove"
      ],
      "modalities": [
        "Whole-hand pressure map",
        "Hand kinematics",
        "Object 6D pose",
        "Language description",
        "Task-phase annotation"
      ],
      "sampleCount": "Not stated on the reviewed project page",
      "tasks": [
        "Single-hand dexterous manipulation",
        "Bimanual manipulation",
        "Human-to-robot skill transfer"
      ],
      "objectCategories": "Household and tool objects shown in project demonstrations; total object count not stated.",
      "dataFormat": "Synchronized tactile pressure, hand-object kinematics, 6D pose, language, and hierarchical task annotations.",
      "license": "Dataset license not stated on the reviewed project page",
      "paperUrl": "https://arxiv.org/abs/2607.09190",
      "projectUrl": "https://tactidex.github.io/",
      "availability": "The official project page documents the benchmark and demonstrations; no separate dataset download URL was verified.",
      "sourceReviewed": "2026-08-16",
      "evidence": {
        "access": "unknown",
        "accessCheckedAt": "2026-08-16",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Synchronized tactile pressure, hand-object kinematics, 6D pose, language, and hierarchical task annotations.",
        "sources": [
          "https://tactidex.github.io/"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Single-hand dexterous manipulation; Bimanual manipulation; Human-to-robot skill transfer. Collection: Human demonstration capture; Bimanual Franka Inspire deployment.",
        "useLimitations": "Human demonstration and hand-object annotations require retargeting for robot control. Download availability and dataset-file terms must be established separately.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-tactidex"
    },
    {
      "id": "freetacman",
      "name": "FreeTacMan",
      "institution": [
        "Shanghai Innovation Institute",
        "The University of Hong Kong",
        "Shanghai Jiao Tong University",
        "Fudan University"
      ],
      "year": 2025,
      "robot": [
        "Robot-free wearable gripper",
        "Piper",
        "Franka"
      ],
      "sensor": [
        "Modular LED-based visuo-tactile sensor"
      ],
      "modalities": [
        "Wrist RGB video",
        "Visuo-tactile video",
        "Tool-center-point pose",
        "Gripper distance"
      ],
      "sampleCount": "More than 3M visuo-tactile image pairs; more than 10K trajectories; 50 tasks",
      "tasks": [
        "Contact-rich demonstrations",
        "Imitation learning",
        "Tactile pretraining",
        "Visuo-tactile manipulation"
      ],
      "objectCategories": "Fifty contact-rich task categories; examples include fragile handling, insertion, stamping, texture classification, and calligraphy.",
      "dataFormat": "MP4 wrist and tactile videos plus timestamped trajectory files containing TCP pose, quaternion, Euler angles, and gripper distance.",
      "license": "MIT in the provider dataset card. The code repository root LICENSE and README identify Apache-2.0 separately from the dataset card. Do not generalize either label to all artifacts or third-party dependencies.",
      "codeLicense": "Apache-2.0 in the repository root LICENSE; the README agrees and qualifies this as unless specified otherwise. Documentation reviewed November 17, 2026; no code execution.",
      "licenseUrl": "https://opensource.org/license/mit",
      "paperUrl": "https://arxiv.org/abs/2506.01941",
      "projectUrl": "https://opendrivelab.com/FreeTacMan",
      "githubUrl": "https://github.com/OpenDriveLab/FreeTacMan",
      "datasetUrl": "https://huggingface.co/datasets/OpenDriveLab/FreeTacMan",
      "availability": "The official project page links the dataset, code, hardware guide, and mirror.",
      "sourceReviewed": "2026-08-16",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-09-13",
        "licenseStatus": "documented",
        "dataLicense": "mit in the provider dataset-card metadata at the pinned revision; individual asset terms were not inspected.",
        "publicFileScale": "24,915 paths in the provider file manifest, including metadata. Bytes, frames and episodes were not independently recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "MP4 wrist and tactile videos plus timestamped trajectory files containing TCP pose, quaternion, Euler angles, and gripper distance.",
        "sources": [
          "https://huggingface.co/datasets/OpenDriveLab/FreeTacMan/tree/030316fb41d6fa1e58cccb4bbe0f6fddbb932671"
        ],
        "revision": "030316fb41d6fa1e58cccb4bbe0f6fddbb932671",
        "listedFileCount": 24915,
        "dataLicenseUrl": "https://opensource.org/license/mit"
      },
      "selectionGuidance": {
        "intendedUse": "Contact-rich demonstrations; Imitation learning; Tactile pretraining; Visuo-tactile manipulation. Collection: Robot-free wearable gripper; Piper; Franka.",
        "useLimitations": "Human-collected tactile demonstrations need an explicit mapping to the target robot, action space and sensor. Collection convenience is not transfer evidence.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-freetacman"
    },
    {
      "id": "humanoid-vta",
      "name": "Humanoid Visual-Tactile-Action Dataset",
      "institution": [
        "Gwangju Institute of Science and Technology"
      ],
      "year": 2025,
      "robot": [
        "Humanoid teleoperation platform",
        "Two Inspire RH56-DFX dexterous hands"
      ],
      "sensor": [
        "1,062 tactile sensors per hand",
        "Piezoresistive tactile carpet"
      ],
      "modalities": [
        "Egocentric RGB",
        "Third-person RGB",
        "Dense tactile pressure",
        "Arm and finger proprioception",
        "Robot action",
        "External pressure heatmap"
      ],
      "sampleCount": "101.9K synchronized samples; approximately 77-80 episodes per task",
      "tasks": [
        "Towel strong pressure",
        "Towel weak pressure",
        "Sponge strong pressure",
        "Sponge weak pressure"
      ],
      "objectCategories": "Two deformable soft objects (towel and sponge), with rigid-object comparison data described in the paper.",
      "dataFormat": "Synchronized visual, 2,124-channel hand tactile, proprioceptive, action, and external pressure signals; public file format not stated.",
      "license": "Dataset access terms not stated on the reviewed paper page",
      "paperUrl": "https://arxiv.org/abs/2510.25725",
      "availability": "The paper describes the dataset; no official public download URL was verified on 2026-08-16.",
      "sourceReviewed": "2026-08-16",
      "evidence": {
        "access": "unknown",
        "accessCheckedAt": "2026-08-16",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Synchronized visual, 2,124-channel hand tactile, proprioceptive, action, and external pressure signals; public file format not stated.",
        "sources": [
          "https://arxiv.org/abs/2510.25725"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Towel strong pressure; Towel weak pressure; Sponge strong pressure; Sponge weak pressure. Collection: Humanoid teleoperation platform; Two Inspire RH56-DFX dexterous hands.",
        "useLimitations": "Full-hand coverage is tied to the collection embodiment. Pressure maps do not establish calibrated shear, fingertip geometry or another hand’s control performance.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-humanoid-vta"
    },
    {
      "id": "sparsh-x",
      "name": "Sparsh-X Multisensory Touch Resource",
      "institution": [
        "FAIR at Meta",
        "University of Washington",
        "Carnegie Mellon University"
      ],
      "year": 2025,
      "robot": [
        "Robot manipulation platforms used for insertion and in-hand rotation"
      ],
      "sensor": [
        "Digit 360"
      ],
      "modalities": [
        "Tactile image",
        "Audio",
        "Motion",
        "Pressure"
      ],
      "sampleCount": "Approximately 1M unlabeled contact-rich interactions",
      "tasks": [
        "Physical-property inference",
        "Plug insertion",
        "In-hand rotation",
        "Tactile adaptation"
      ],
      "objectCategories": "Diverse manipulation interactions; object count and category inventory not stated on the reviewed paper page.",
      "dataFormat": "Four synchronized Digit 360 modalities; public file format not stated on the reviewed paper page.",
      "license": "Dataset license and download URL not stated on the reviewed paper page",
      "paperUrl": "https://arxiv.org/abs/2506.14754",
      "availability": "The paper documents the training resource; a dedicated public dataset URL was not verified.",
      "sourceReviewed": "2026-08-16",
      "evidence": {
        "access": "unknown",
        "accessCheckedAt": "2026-08-16",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Four synchronized Digit 360 modalities; public file format not stated on the reviewed paper page.",
        "sources": [
          "https://arxiv.org/abs/2506.14754"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Physical-property inference; Plug insertion; In-hand rotation; Tactile adaptation. Collection: Robot manipulation platforms used for insertion and in-hand rotation.",
        "useLimitations": "Check which data, encoders and sensor modalities are included in the selected release. A multisensory resource does not guarantee a shared schema or cross-sensor accuracy.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-sparsh-x"
    },
    {
      "id": "touch-and-go",
      "name": "Touch and Go",
      "institution": [
        "University of Michigan",
        "Carnegie Mellon University"
      ],
      "year": 2022,
      "robot": [
        "Human-operated GelSight collection device"
      ],
      "sensor": [
        "GelSight"
      ],
      "modalities": [
        "Egocentric RGB video",
        "GelSight tactile video",
        "Material labels"
      ],
      "sampleCount": "Approximately 13.9K detected touches; approximately 3,971 object instances; 20 material categories",
      "tasks": [
        "Visuo-tactile feature learning",
        "Material recognition",
        "Tactile-driven image stylization",
        "Future tactile prediction"
      ],
      "objectCategories": "Rigid and deformable objects across indoor and outdoor environments; the paper reports 20 material categories.",
      "dataFormat": "Synchronized egocentric video and GelSight tactile recordings with detected-touch timing and material annotations.",
      "license": "Dataset license not stated on the reviewed paper page",
      "paperUrl": "https://arxiv.org/abs/2211.12498",
      "projectUrl": "https://touch-and-go.github.io/",
      "availability": "The paper and project describe a human-collected in-the-wild visuo-tactile dataset; verify the current download terms before reuse.",
      "sourceReviewed": "2026-08-19",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-08-19",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Synchronized egocentric video and GelSight tactile recordings with detected-touch timing and material annotations.",
        "sources": [
          "https://touch-and-go.github.io/"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Visuo-tactile feature learning; Material recognition; Tactile-driven image stylization; Future tactile prediction. Collection: Human-operated GelSight collection device.",
        "useLimitations": "Use aligned human-collected visual and tactile observations for perception. They do not provide a standardized robot-action or closed-loop slip benchmark.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-touch-and-go"
    },
    {
      "id": "tvl",
      "name": "TVL: Touch, Vision, and Language",
      "institution": [
        "UC Berkeley",
        "Meta AI Research",
        "TU Dresden",
        "CeTI"
      ],
      "year": 2024,
      "robot": [
        "Handheld 3D-printed HCT collection device",
        "Robot-collected SSVTP subset"
      ],
      "sensor": [
        "DIGIT"
      ],
      "modalities": [
        "Tactile RGB image",
        "RGB image",
        "English tactile description"
      ],
      "sampleCount": "43,741 in-contact image-touch pairs; HCT contributes 39,154 pairs and SSVTP contributes 4,587 pairs",
      "tasks": [
        "Tactile-semantic classification",
        "Touch-vision alignment",
        "Touch-language alignment",
        "Tactile description generation"
      ],
      "objectCategories": "In-the-wild surfaces and objects plus the SSVTP laboratory subset; a total object-category inventory is not stated on the reviewed project page.",
      "dataFormat": "Temporally aligned touch, vision, and open-vocabulary language examples; 10% of labels are human annotations and 90% are VLM-generated in the paper.",
      "license": "Dataset license not stated on the reviewed project page",
      "paperUrl": "https://arxiv.org/abs/2402.13232",
      "projectUrl": "https://tactile-vlm.github.io/",
      "datasetUrl": "https://huggingface.co/datasets/yoorhim/TVL-revise/tree/aab39e037d8916717dc98add89d58cbd39219770",
      "datasetUrlReviewed": "2026-11-15",
      "availability": "The official repository links TVL-revise and reports that it fixes swapped tactile/image folders in the original upload. The revised host manifest is public; the correction has not been independently validated by loading image pairs. Dataset-file reuse terms remain unverified.",
      "sourceReviewed": "2026-08-19",
      "evidence": {
        "access": "public-files",
        "accessCheckedAt": "2026-11-15",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: the inspected TVL-revise card and host metadata do not specify dataset-file terms. The upstream code license is not treated as a dataset license.",
        "publicFileScale": "10 paths in the pinned TVL-revise host manifest: 8 split-archive parts, README.md and .gitattributes. Images, frames and bytes were not independently recounted.",
        "verificationScope": "Upstream correction notice, revised dataset card and public host manifest inspected at pinned revisions. The folder-swap fix is source-reported; no archive download, extraction, image-pair validation, loader execution or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Temporally aligned touch, vision, and open-vocabulary language examples; 10% of labels are human annotations and 90% are VLM-generated in the paper.",
        "revision": "aab39e037d8916717dc98add89d58cbd39219770",
        "listedFileCount": 10,
        "sources": [
          "https://tactile-vlm.github.io/",
          "https://github.com/Max-Fu/tvl/blob/f489503d829be1eec1f800070cdbec02aede4246/README.md#improved-version-available",
          "https://huggingface.co/datasets/yoorhim/TVL-revise/tree/aab39e037d8916717dc98add89d58cbd39219770",
          "https://huggingface.co/datasets/mlfu7/Touch-Vision-Language-Dataset"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Tactile-semantic classification; Touch-vision alignment; Touch-language alignment; Tactile description generation. Collection: Handheld 3D-printed HCT collection device; Robot-collected SSVTP subset.",
        "useLimitations": "Use touch, vision and language for multimodal representations. Language alignment does not establish manipulation success, force calibration or action supervision.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-tvl"
    },
    {
      "id": "objectfolder-real",
      "name": "ObjectFolder Real",
      "institution": [
        "Stanford University",
        "Carnegie Mellon University"
      ],
      "year": 2023,
      "robot": [
        "Franka Emika Panda"
      ],
      "sensor": [
        "GelSight robotic finger"
      ],
      "modalities": [
        "3D mesh",
        "HD RGB video",
        "Impact audio",
        "Tactile reading"
      ],
      "sampleCount": "100 real-world household objects",
      "tasks": [
        "Cross-sensory retrieval",
        "Contact localization",
        "Material classification",
        "Reconstruction",
        "Manipulation benchmarks"
      ],
      "objectCategories": "One hundred real-world household objects; detailed inventory is provided through the official ObjectFolder resources.",
      "dataFormat": "Multisensory measurements aligned to object surface locations, including mesh, visual, acoustic, and tactile assets.",
      "license": "Review the license bundled with the downloaded ObjectFolder release",
      "paperUrl": "https://openaccess.thecvf.com/content/CVPR2023/html/Gao_The_ObjectFolder_Benchmark_Multisensory_Learning_With_Neural_and_Real_Objects_CVPR_2023_paper.html",
      "projectUrl": "https://objectfolder.stanford.edu/",
      "githubUrl": "https://github.com/rhgao/ObjectFolder",
      "availability": "The official ObjectFolder site provides dataset and benchmark download routes.",
      "sourceReviewed": "2026-08-19",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-08-19",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Multisensory measurements aligned to object surface locations, including mesh, visual, acoustic, and tactile assets.",
        "sources": [
          "https://objectfolder.stanford.edu/"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Cross-sensory retrieval; Contact localization; Material classification; Reconstruction; Manipulation benchmarks. Collection: Franka Emika Panda.",
        "useLimitations": "Object-level real observations support multisensory perception. A fixed object collection is not a long-horizon robot-manipulation dataset.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-objectfolder-real"
    },
    {
      "id": "objectfolder-2",
      "name": "ObjectFolder 2.0",
      "institution": [
        "Stanford University",
        "Carnegie Mellon University"
      ],
      "year": 2022,
      "robot": [
        "Simulation / neural object rendering"
      ],
      "sensor": [
        "Simulated GelSight-style tactile rendering"
      ],
      "modalities": [
        "Visual appearance",
        "Impact audio",
        "Tactile RGB image",
        "3D geometry"
      ],
      "sampleCount": "1,000 multisensory neural objects",
      "tasks": [
        "Object scale estimation",
        "Contact localization",
        "Shape reconstruction",
        "Multisensory representation learning"
      ],
      "objectCategories": "One thousand neural objects derived from real-object 3D assets across common household categories.",
      "dataFormat": "Object File implicit neural representations with query parameters for view, lighting, impact, contact location, gel rotation, and indentation depth.",
      "license": "CC BY 4.0 for ObjectFolder 2.0 according to the official download page",
      "licenseUrl": "https://creativecommons.org/licenses/by/4.0/",
      "paperUrl": "https://arxiv.org/abs/2204.02389",
      "projectUrl": "https://objectfolder.stanford.edu/objectfolder2-0-download",
      "githubUrl": "https://github.com/rhgao/ObjectFolder",
      "datasetUrl": "https://objectfolder.stanford.edu/objectfolder2-0-download",
      "availability": "The official download page provides the Object Files, rendering instructions, metadata, and license information.",
      "sourceReviewed": "2026-08-19",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-08-19",
        "licenseStatus": "documented",
        "dataLicense": "CC BY 4.0 for ObjectFolder 2.0 according to the existing review of the official download page.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Object File implicit neural representations with query parameters for view, lighting, impact, contact location, gel rotation, and indentation depth.",
        "sources": [
          "https://objectfolder.stanford.edu/objectfolder2-0-download"
        ],
        "dataLicenseUrl": "https://creativecommons.org/licenses/by/4.0/"
      },
      "selectionGuidance": {
        "intendedUse": "Object scale estimation; Contact localization; Shape reconstruction; Multisensory representation learning. Collection: Simulation / neural object rendering.",
        "useLimitations": "Simulated object signals support controlled representation experiments. Rendering fidelity and transfer to real sensors require separate tests.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-objectfolder-2"
    },
    {
      "id": "tacverse",
      "name": "TacVerse",
      "institution": [
        "Imperial College London",
        "Queen Mary University of London",
        "King’s College London"
      ],
      "year": 2026,
      "robot": [
        "Controlled tactile data-collection platform"
      ],
      "sensor": [
        "GelSightNoMarker",
        "GelSightMarker",
        "MagicGripper",
        "MagicTac",
        "TacTip",
        "ViTac",
        "ViTacTip"
      ],
      "modalities": [
        "Tactile RGB image",
        "Shape label",
        "Grating label",
        "Force label"
      ],
      "sampleCount": "106,800 tactile images from seven vision-based tactile sensors",
      "tasks": [
        "Shape classification",
        "Grating classification",
        "Force regression",
        "Zero-shot cross-sensor transfer",
        "Few-shot adaptation"
      ],
      "objectCategories": "Controlled shape and grating stimuli plus force-regression contacts; a general object-category inventory is not stated.",
      "dataFormat": "Multi-sensor tactile images and task labels organized for within-sensor, zero-shot cross-sensor, and few-shot protocols.",
      "license": "CC BY 4.0 in the official Hugging Face metadata; access is gated and asset-specific terms were not independently checked",
      "paperUrl": "https://arxiv.org/abs/2606.25877",
      "projectUrl": "https://lannwei.github.io/Tactile_Database/",
      "githubUrl": "https://github.com/LannWei/Tactile_Database",
      "datasetUrl": "https://huggingface.co/datasets/Lan-2025/Tactile",
      "researchUrl": "/research/tacverse-cross-sensor-tactile-dataset-2026",
      "availability": "Official code and gated Hugging Face task archives verified on 2026-09-11. The 106,800-image count is paper-reported; files were not downloaded or recounted. No root code license was identified at the reviewed revision.",
      "sourceReviewed": "2026-09-11",
      "evidence": {
        "access": "gated",
        "accessCheckedAt": "2026-09-13",
        "licenseStatus": "documented",
        "dataLicense": "cc-by-4.0 in the provider dataset-card metadata at the pinned revision; individual asset terms were not inspected.",
        "publicFileScale": "7 paths in the provider file manifest, including metadata. Bytes, frames and episodes were not independently recounted.",
        "verificationScope": "Official hosting API and file manifest inspected at the pinned revision. No dataset payloads downloaded; no checksums, schema validation or training run.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Multi-sensor tactile images and task labels organized for within-sensor, zero-shot cross-sensor, and few-shot protocols.",
        "sources": [
          "https://huggingface.co/datasets/Lan-2025/Tactile/tree/0bc27afe0d8f6c878b79e1eb0825255541ccaceb"
        ],
        "revision": "0bc27afe0d8f6c878b79e1eb0825255541ccaceb",
        "listedFileCount": 7,
        "dataLicenseUrl": "https://creativecommons.org/licenses/by/4.0/"
      },
      "selectionGuidance": {
        "intendedUse": "Shape classification; Grating classification; Force regression; Zero-shot cross-sensor transfer; Few-shot adaptation. Collection: Controlled tactile data-collection platform.",
        "useLimitations": "Cross-sensor comparisons require the documented sensor and task splits. Do not compare accuracy numbers across different split or few-shot settings.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-tacverse"
    },
    {
      "id": "vtdexmanip",
      "name": "VTDexManip",
      "institution": [
        "Zhejiang University"
      ],
      "year": 2025,
      "robot": [
        "Human dexterous demonstration platform",
        "Simulated dexterous manipulation benchmark"
      ],
      "sensor": [
        "Sparse binary tactile sensing"
      ],
      "modalities": [
        "RGB image",
        "Sparse binary touch",
        "Proprioception",
        "Manipulation trajectory"
      ],
      "sampleCount": "10 daily manipulation tasks across 182 objects; trajectory and frame totals are not stated as text on the reviewed project page",
      "tasks": [
        "Visual-tactile pretraining",
        "Bottle-cap turning",
        "Faucet screwing",
        "Lever sliding",
        "Table reorientation",
        "In-hand reorientation",
        "Bimanual hand-over"
      ],
      "objectCategories": "One hundred eighty-two objects used across ten daily manipulation tasks.",
      "dataFormat": "Visual-tactile pretraining data and an Isaac Gym policy benchmark; inspect the official repository for the current file layout.",
      "license": "MIT License for the benchmark code; verify dataset-file terms separately",
      "licenseUrl": "https://opensource.org/license/mit",
      "paperUrl": "https://openreview.net/forum?id=jf7C7EGw21",
      "projectUrl": "https://lqts.github.io/VTDexManip/",
      "githubUrl": "https://github.com/LQTS/VTDexManip",
      "availability": "The author project and repository link benchmark code, paper, and dataset access.",
      "sourceReviewed": "2026-08-19",
      "evidence": {
        "access": "linked-unverified",
        "accessCheckedAt": "2026-08-19",
        "licenseStatus": "unknown",
        "dataLicense": "Unknown: a dataset-file license has not been verified separately from paper or code terms.",
        "publicFileScale": "Unknown: no independently recounted public-file total in this record.",
        "verificationScope": "Existing source review only. No new dataset-file download, checksum validation, training run, or hardware test in this audit.",
        "splitStatus": "unknown",
        "splitDetails": "Unknown: no train/test split manifest verified in this audit.",
        "formatStatus": "described",
        "formatDetails": "Visual-tactile pretraining data and an Isaac Gym policy benchmark; inspect the official repository for the current file layout.",
        "sources": [
          "https://lqts.github.io/VTDexManip/"
        ]
      },
      "selectionGuidance": {
        "intendedUse": "Visual-tactile pretraining; Bottle-cap turning; Faucet screwing; Lever sliding; Table reorientation; In-hand reorientation; Bimanual hand-over. Collection: Human dexterous demonstration platform; Simulated dexterous manipulation benchmark.",
        "useLimitations": "Separate visual-tactile pretraining from the simulator policy benchmark. Neither alone demonstrates transfer to a physical hand or a new sensor.",
        "basis": "Editorial guidance from the linked source record; not an independent reproduction."
      },
      "directoryRecordUrl": "https://roboskin.ai/datasets#dataset-vtdexmanip"
    }
  ]
}
