Geppetto tracks 1035 robot training datasets including Open X-Embodiment, DROID, and AgiBot World, each with structured metadata on trajectory count, hours of demonstration, robot platforms, license, commercial use, access model, and file format. Every dataset connects to the robots it trains and the professions it affects.
| Dataset | Year | Category | Trajectories | Hours | Platforms | License | Commercial | Access | Format |
|---|---|---|---|---|---|---|---|---|---|
| ABC-130k | 2026 | real-robot | — | 3,598.2 | parallel_jaw, aloha | Apache-2.0 | Yes | open | — |
| LeRobot Shirt-Folding Dataset | 2026 | real-robot | — | 131 | parallel_jaw | Apache-2.0 | Yes | open | — |
| EgoInfinity | 2026 | human-video | — | — | franka_panda, human, unitree_humanoid | custom | No | open | — |
| LESS (Local Encoder for Spatial Sensing) | 2026 | human-video | — | 800 | franka_panda | CC-BY-4.0 | Yes | open | — |
| FTP-1 Dataset | 2026 | real-robot | — | 3,000 | multi | MIT | Yes | open | — |
| MolmoAct Dataset | 2026 | real-robot | — | 22 | franka_panda | Apache-2.0 | Yes | open | — |
| MolmoAct2 SO-100/SO-101 Dataset | 2026 | real-robot | — | — | so_100, so_101 | Apache-2.0 | Yes | open | — |
| MolmoAct2 Bimanual YAM Dataset | 2026 | real-robot | — | 720 | yam | Apache-2.0 | Yes | open | — |
| ManipArena | 2026 | real-robot | — | 188 | parallel_jaw, aloha | Apache-2.0 | Yes | open | — |
| MolmoBot | 2026 | synthetic | — | 5,704 | franka_panda | custom | No | open | — |
| DreamDojo GR-1 Post-Training | 2026 | human-video | — | 88.4 | humanoid_other | CC-BY-NC-4.0 | No | open | — |
| HiFi-UMI-2K | 2026 | human-video | — | 2,137 | human, umi_gripper | CC-BY-4.0 | Yes | open | — |
| Xiaomi XR-1 and XR-0 post-train demos | 2026 | real-robot | — | — | Xiaomi robot demo episodes | Apache-2.0 | Yes | open | — |
| DuoBench | 2026 | real-robot | — | — | Franka FR3 Duo | Apache-2.0 | Yes | open | — |
| GigaWorld-1 toy data (WMBench) | 2026 | research-lab | — | — | not stated on opened page | Apache-2.0 | Yes | open | — |
| EgoHumanoid sample | 2026 | real-robot | — | — | Unitree G1 teleop plus PICO VR and ZED human demos | apache-2.0 | Yes | open | — |
| GigaBrain-0.7 Sample Data | 2026 | real-robot | — | — | GigaAI robot sample episodes | Apache-2.0 | Yes | open | — |
| RynnBrain-Bench | 2026 | research-lab | — | — | embodied video | apache-2.0 | Yes | open | — |
| ASIG-Lab/JEPLO-Dataset | 2026 | research-lab | — | — | not stated | gpl-3.0 | Yes | open | — |
| Humanola egocentric sample | 2026 | research-lab | — | — | First-person video | CC BY 4.0 | Yes | open | — |
| Hub Egocentric Human Demonstrations Sample Set | 2026 | human-video | — | — | GoPro HERO13, iPhone, ZED X Mini | other | No | open | — |
| EgoSuite-Open100K: EgoStandard | 2026 | human-video | — | — | Head camera, hand and body pose | custom (other) | No | gated | — |
| BeingBeyond/UniHand_Preview | 2026 | research-lab | — | — | not stated | not stated | No | open | — |
| ERQA+ (ERQA Plus) | 2026 | research-lab | — | — | embodied reasoning images | mit | Yes | open | — |
| Chuhaojin/SuSuInterActs | 2026 | research-lab | — | — | not stated | cc-by-4.0 | Yes | open | — |
| PRISM data (Amazon FAR) | 2026 | synthetic | — | — | Unitree G1 loco-manipulation | apache-2.0 | Yes | open | — |
| MobileVLA-CoT | 2026 | research-lab | — | — | quadruped robots | not stated | No | open | — |
| VoLN-UAV | 2026 | synthetic | — | — | simulated UAV | MIT | Yes | open | — |
| Extreme Driving Dataset V2 | 2026 | research-lab | — | — | vehicle with 6 surround cameras and ZED stereo | CC BY-NC 4.0 | No | gated | — |
| ZereData Bin Picking Dataset v1.2 | 2026 | research-lab | — | — | synthetic bin scenes, RGB, depth, masks, 6D pose | CC BY 4.0 | Yes | open | — |
| 60base Korea household egocentric samples | 2026 | human-video | — | — | Egocentric household video | other | No | open | — |
| PRISM-100K (DreamVu) | 2026 | research-lab | — | — | synchronized ego and exo retail video | cc-by-nc-4.0 | No | gated | — |
| Labelbox robotics datasets (LBX preview) | 2026 | real-robot | — | — | Two bimanual xArm stations, raw MCAP | CC BY 4.0 | Yes | open | — |
| Go2 Air ControlBench v1 | 2026 | research-lab | — | — | Unitree Go2 Air | CC BY 4.0 | Yes | open | — |
| ReMind1M | 2026 | research-lab | — | — | Captioned video with trajectory profiles | custom (other) | No | open | — |
| LIBERO-Para | 2026 | research-lab | — | — | Franka Panda in LIBERO | mit | Yes | open | — |
| Airborne Object Tracking (AOT) | 2026 | research-lab | — | — | aircraft-mounted cameras | not stated | No | open | — |
| ForVis | 2026 | research-lab | — | — | under-canopy UAV with RealSense D435i and OAK-D Pro Wide | not stated | No | open | — |
| LARD 2.0 | 2026 | research-lab | — | — | aircraft landing approach imagery | not stated | No | open | — |
| NanoBench | 2026 | research-lab | — | — | nano-quadrotor | not stated | No | open | — |
| KAIST Multispectral Pedestrian Dataset | 2026 | research-lab | — | — | RGB and thermal cameras | not stated | No | open | — |
| BiGS (BioTac Grasp Stability) | 2026 | research-lab | — | — | BioTac sensors on robot hand | not stated | No | open | — |
| GRAB food waste grasping-in-clutter benchmark | 2026 | research-lab | — | — | industrial grasping, three gripper modalities | not stated | No | open | — |
| SCT-CNN grasp stability dataset | 2026 | research-lab | — | — | not confirmed | not stated | No | open | — |
| uSkin Slip Dataset | 2026 | research-lab | — | — | not confirmed | not stated | No | open | — |
| Generalist AI pretraining corpus (GEN-1) | 2026 | research-lab | — | — | wearable and handheld data collection devices ("robot hands"); the GEN-1 post says the pretraining set contains no robot data | not stated | No | open | — |
| Grabette community data (pollen-robotics/grabette-community-tasks and related) | 2026 | research-lab | — | — | Grabette handheld gripper (fisheye and RGB-D cameras, IMU) | not stated | No | open | — |
| TRUMI datasets (TrossenRoboticsCommunity/trumi-dataset, example_trumi_dataset) | 2026 | research-lab | — | — | Trossen UMI-style handheld gripper (inferred from name; repos have no card) | not stated | No | open | — |
| UMI-Underwater | 2026 | research-lab | — | — | underwater gripper system plus on-land handheld UMI | not stated | No | open | — |
| Helix 02 System 0 human motion data | 2026 | research-lab | — | — | human motion retargeted to Figure humanoid joints | not stated | No | open | — |
| Humanoid-GPT retargeted corpus | 2026 | research-lab | — | — | humanoid robot | not stated | No | open | — |
| ACCAD motion capture data | 2026 | research-lab | — | — | human body, Vicon | not stated | No | open | — |
| Bandai Namco Research Motion Dataset 1 and 2 | 2026 | research-lab | — | — | human body, BVH skeleton | not stated in excerpt | No | open | — |
| CMU Graphics Lab Motion Capture Database | 2026 | research-lab | — | — | human body, Vicon marker mocap | not stated | No | open | — |
| Dance2Hesitate | 2026 | research-lab | — | — | dancer motion and kinesthetic robot teaching | not stated | No | open | — |
| HDM05 | 2026 | research-lab | — | — | human body, marker mocap | not stated | No | open | — |
| Human3.6M | 2026 | research-lab | — | — | human body, 4 calibrated cameras plus mocap | custom licence page, not read | No | open | — |
| KIT Whole-Body Human Motion Database | 2026 | research-lab | — | — | human body plus manipulated objects, Master Motor Map reference model | not stated | No | open | — |
| SFU Motion Capture Database | 2026 | research-lab | — | — | human body, BVH | not stated | No | open | — |
| CMU Panoptic Dataset | 2026 | research-lab | — | — | multi-camera dome | not stated | No | open | — |
| OHRID-Retail | 2026 | research-lab | — | — | multimodal sensors on humans, shelf picking | not stated | No | open | — |
| AirNav | 2026 | research-lab | — | — | real urban aerial data | not stated | No | open | — |
| UAVBench and UAVIT-1M | 2026 | research-lab | — | — | low-altitude UAV imagery | not stated | No | open | — |
| FailBench | 2026 | research-lab | — | — | 14 public sources, 12 real-world and 2 simulated | not stated | No | open | — |
| HumanoidVLN | 2026 | research-lab | — | — | Isaac Sim humanoids incl. Unitree G1 and H1 | not stated | No | open | — |
| CapNav | 2026 | research-lab | — | — | real indoor scenes | not stated | No | open | — |
| EQA-Decision | 2026 | research-lab | — | — | embodied scenarios | not stated | No | open | — |
| A2A-Bench (Affordance2Action) | 2026 | research-lab | — | — | everyday scene images | not stated | No | open | — |
| FG-CLTP dataset | 2026 | research-lab | — | — | tactile 3D point clouds | not stated | No | open | — |
| TouchCoT-10k | 2026 | research-lab | — | — | tactile inputs with chain-of-thought text | not stated | No | open | — |
| TouchReason-1M | 2026 | research-lab | — | — | four tactile sensors | not stated | No | open | — |
| TouchThinker-1M | 2026 | research-lab | — | — | 7 tactile sensor types | not stated | No | open | — |
| DaViNCi | 2026 | research-lab | — | — | outdoor continuous simulator with dynamic elements | not stated | No | open | — |
| DPed-VLN | 2026 | research-lab | — | — | Habitat 3.0 with humanoid pedestrians | not stated | No | open | — |
| IntentionNav | 2026 | research-lab | — | — | Isaac Sim scenes | not stated | No | open | — |
| MAVLN | 2026 | research-lab | — | — | multi-agent simulated teams | not stated | No | open | — |
| POINav-Dataset | 2026 | research-lab | — | — | 3DGS reconstructions of commercial areas | not stated | No | open | — |
| Rule-VLN | 2026 | research-lab | — | — | Touchdown / StreetLearn environment | other | No | open | — |
| Urban airspace goal navigation benchmark | 2026 | research-lab | — | — | UAV in urban airspace | not stated | No | open | — |
| RCT | 2026 | research-lab | — | — | robot presses on 122 industrial reference materials | not stated | No | open | — |
| AnchorReasoning | 2026 | research-lab | — | — | WOD-E2E | not stated | No | open | — |
| Drive-P2D | 2026 | research-lab | — | — | driving images | not stated | No | open | — |
| DriveReward | 2026 | research-lab | — | — | driving video with trajectories | not stated | No | open | — |
| NARRATE | 2026 | research-lab | — | — | instrumented car, Australia | not stated | No | open | — |
| NeuroSymbEAD | 2026 | research-lab | — | — | KITTI-360 | not stated | No | open | — |
| V2X-QA | 2026 | research-lab | — | — | vehicle, infrastructure and cooperative views | not stated | No | open | — |
| Agile perceptive multi-skill quadruped locomotion raw data | 2026 | research-lab | — | — | quadrupedal robot | CC BY 4.0 | Yes | open | — |
| M-SEVIQ | 2026 | research-lab | — | — | Unitree Go2 with stereo event cameras | not stated | No | open | — |
| OCELOT dataset | 2026 | research-lab | — | — | legged robot, multi-modal | not stated | No | open | — |
| QuadFM | 2026 | research-lab | — | — | quadruped robot | not stated | No | open | — |
| WHU-PCPR | 2026 | research-lab | — | — | vehicle-mounted and helmet-mounted laser scanners | not stated | No | open | — |
| GuideTWSI | 2026 | research-lab | — | — | tactile paving images, guide robots | not stated | No | open | — |
| Mag4D-SLAM | 2026 | research-lab | — | — | LiDAR, camera, IMU, tri-axis magnetometer | not stated | No | open | — |
| MPO multi-modal panoramic 3D outdoor datasets | 2026 | research-lab | — | — | panoramic 3D lidar scanners | not stated | No | open | — |
| ScaleMaster | 2026 | research-lab | — | — | monocular camera in large indoor environments | not stated | No | open | — |
| RoboVAST dataset | 2026 | research-lab | — | — | simulated mobile robot, model not stated | not stated | No | open | — |
| Bi3 | 2026 | research-lab | — | — | two robot platforms in a lab space | not stated | No | open | — |
| eNavi | 2026 | real-robot | — | — | TurtleBot 2 with event camera and RGB camera | not stated | No | open | — |
| Lincoln Long-Term Mobile Robot Operations datasets | 2026 | research-lab | — | — | not stated | not stated | No | open | — |
| MIT Stata Center Data Set | 2026 | research-lab | — | — | not stated | not stated | No | open | — |