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 |
|---|---|---|---|---|---|---|---|---|---|
| Humanity's Last Machine — Humanoid Robot Supplier Database | 2026 | supply-chain-intelligence | — | — | — | All rights reserved — data sourced from public stock exchange filings and company announcements. Attribution required. | No | open | csv |
| GrandTour — Legged Robot Dataset in the Wild | 2026 | real-robot | — | — | ANYbotics ANYmal-D | MIT — confirm in repository LICENSE file before activation | Yes | gated | rosbag |
| PhysicalAI Robotics GR00T X-Embodiment Sim | 2026 | synthetic | — | — | Franka Panda, Isaac Lab robots, humanoid robots | CC BY 4.0 | Yes | open | parquet, zarr |
| OmniAction — Multimodal Proactive Robot Manipulation Dataset | 2026 | real-robot | — | — | robot arms, humanoid arms | CC BY-NC 4.0 | No | open | mp4, parquet, json |
| Xperience-10M — Large-Scale Egocentric Multimodal Human Experience Dataset | 2026 | human-video | — | 10,000 | human (wearable camera), egocentric rig | Apache 2.0 | Yes | gated | mp4, json, hdf5 |
| Egocentric-100K — Manual Labor Egocentric Video Dataset | 2026 | human-video | — | 100,405 | human (head-mounted fisheye camera) | Apache 2.0 | Yes | gated | mp4, json |
| Galaxea Open-World Dataset — Mobile Manipulation | 2026 | real-robot | — | 500 | Galaxea R1 dual-arm mobile robot | CC BY-NC-SA 4.0 | No | open | mp4, parquet, json |
| PhysicalAI Robotics Kitchen Manipulation Demos | 2026 | real-robot | — | 600 | Franka Panda, Omron mobile base | CC BY 4.0 | Yes | open | hdf5, parquet, mp4 |
| GR1 Robot Teleoperation Dataset | 2026 | real-robot | — | — | Fourier GR-1, Fourier GR-1T1 | MIT | Yes | open | hdf5, mp4 |
| PalmDex — Embodiment-Agnostic Tactile Manipulation Dataset | 2026 | real-robot | — | — | human hands, dexterous robot hands | CC BY-SA 4.0 | Yes | open | mp4, hdf5, json |
| Egocentric-1M — The Internet for Physical AI | 2026 | human-video | — | 1,000,000 | human (head-mounted fisheye camera) | Apache 2.0 | Yes | gated | mp4, json, webdataset |
| AgiBot World 2026 | 2026 | real-robot | — | 2,976.4 | AgiBot G2 | CC BY-NC-SA 4.0 | No | gated | parquet, mp4, json |
| RoboReward — Vision-Language Reward Dataset for Robotics | 2026 | real-robot | — | — | Open X-Embodiment robots, RoboArena robots | CC BY 4.0 | Yes | open | mp4, parquet, json |
| LingBot-VLA 2.0 Pretraining Dataset | 2026 | real-robot | — | 60,000 | Leju, AgiBot, Unitree | Apache-2.0 | Yes | open | — |
| Industrial Embodied AI Robot Training Dataset 2.0 | 2026 | real-robot | — | — | Estun, Hisense, ZTE | Agreement-required | No | agreement-required | — |
| REAL-I Industrial Humanoid Dataset | 2026 | real-robot | 180,000 | 60,000 | Kuavo 4 Pro, Kuavo 5, Kuavo | Apache-2.0 | Yes | open | — |
| SPARC (Spatial Annotations from Robot Demonstrations at Scale) | 2026 | research-lab | 1,700 | — | Franka, bimanual setups, 12 diverse embodiments | null | No | open | — |
| EgoEngine | 2026 | research-lab | — | — | RB-Y1 Humanoid Robot | MIT | Yes | open | — |
| HiPHI (High-Precision Human Motion Dataset) | 2026 | human-video | — | 617.5 | — | ModalityNet Open Research License v1.0 | No | open | — |
| Axis Sim Dataset V1 | 2026 | synthetic | 50,000 | — | Franka Research 3 | open-source | Yes | open | — |
| RoboEdit-14M | 2026 | real-robot | 174,547 | 130.91 | various | arXiv.org perpetual non-exclusive license | No | open | — |
| Data Pyramid for Embodied Manipulation Survey Dataset | 2026 | research-lab | — | — | — | null | No | open | — |
| EgoLive | 2026 | human-video | — | 1,680 | — | open-source | Yes | open | — |
| RoboDojo | 2026 | research-lab | — | 38.57 | parallel_jaw, aloha | Apache-2.0 (per datasets.bot) | Yes | open | — |
| PanoAir | 2026 | research-lab | — | — | UAV with panoramic camera and IMU | not stated | No | open | — |
| UAV Active Object Detection in the Wild dataset | 2026 | research-lab | — | — | UAV panoramic imagery | not stated | No | open | — |
| UAV3DCrop | 2026 | research-lab | — | — | UAV multi-angle RGB surveys | CC-BY-4.0 | Yes | open | — |
| LRDDv3 | 2026 | research-lab | — | — | long-range RGB and thermal images of drones | not stated | No | open | — |
| SkyDrive | 2026 | research-lab | — | — | traffic monitoring drones | not stated | No | open | — |
| V2U4Real | 2026 | research-lab | — | — | vehicle plus UAV, lidar and cameras | not stated | No | open | — |
| RESOLVE | 2026 | research-lab | — | — | roadside multi-resolution cameras and lidar | not stated | No | open | — |
| uniD | 2026 | research-lab | — | — | drone over RWTH campus shared space | not stated | No | open | — |
| 4DLidarOpen | 2026 | research-lab | — | — | vehicle with forward 4D FMCW lidar, Beijing | not stated | No | open | — |
| A2RL Vmax | 2026 | research-lab | — | — | A2RL race cars at Yas Marina | not stated | No | open | — |
| BATON | 2026 | research-lab | — | — | production vehicles with driving automation | not stated | No | open | — |
| Boreas Road Trip (Boreas-RT) | 2026 | research-lab | — | — | Boreas car plus Aeva Aeries II FMCW lidar, Doppler radar | not stated | No | open | — |
| BranchDrive | 2026 | research-lab | — | — | driving, action-conditioned intervention branches | not stated | No | open | — |
| DarkDriving | 2026 | research-lab | — | — | vehicle cameras on closed test field | not stated | No | open | — |
| DRIVE-C | 2026 | research-lab | — | — | camera clips with 12 degradation types | not stated | No | open | — |
| DrivIng | 2026 | research-lab | — | — | multimodal vehicle with geo-referenced digital twin | not stated | No | open | — |
| eAP | 2026 | research-lab | — | — | event camera vehicle rig | not stated | No | open | — |
| FRED (Flooded Road Environments) | 2026 | research-lab | — | — | FLIR Blackfly camera, Ouster OS1-64, RTK IMU | not stated | No | open | — |
| IEDD (Interactive Enhanced Driving Dataset) | 2026 | research-lab | — | — | ego-centric interaction segments from driving logs | not stated | No | open | — |
| Malaga Stereo and Laser Urban Data Set | 2026 | research-lab | — | — | stereo camera and lasers on car | not stated | No | open | — |
| Multi-Vehicle Dataset with scanned 3D vehicle models (RTK-GNSS auto-annotation) | 2026 | research-lab | — | — | camera, LiDAR, radar vehicles | not stated | No | open | — |
| NVIDIA OmniDreams driving training corpus | 2026 | research-lab | — | — | not stated | not stated | No | open | — |
| OVPD (OnSite Virtual-Physical Dataset) | 2026 | research-lab | — | — | virtual-physical fusion AV testing | not stated | No | open | — |
| ReaLiTy and LADS | 2026 | research-lab | — | — | LiDAR across sensors and adverse weather | not stated | No | open | — |
| SearchAD | 2026 | research-lab | — | — | frames from 11 driving datasets | not stated | No | open | — |
| SinD 2.0 | 2026 | research-lab | — | — | UAV-recorded signalized intersections | not stated | No | open | — |
| SWIFTraj | 2026 | research-lab | — | — | UAV-recorded freeway and urban trajectories | not stated | No | open | — |
| Transitional Autonomous Vehicle Lane-Changing Dataset | 2026 | research-lab | — | — | transitional autonomous vehicle | not stated | No | open | — |
| TuSimple lane and velocity benchmark | 2026 | research-lab | — | — | highway camera | not stated | No | open | — |
| VRUD | 2026 | research-lab | — | — | drone-recorded vehicle and VRU traffic | not stated | No | open | — |
| UniHand-2.0 (Being-H0.5) | 2026 | human-video | — | — | human hands plus 30 robot embodiments | not stated | No | open | — |
| TactiDex | 2026 | human-video | — | — | Human hand with whole-hand tactile sensing; viewpoint not stated in abstract | not stated | No | open | — |
| 1X World Model egocentric human video set (1XWM) | 2026 | human-video | — | — | not stated | not stated | No | open | — |
| AirGlove | 2026 | human-video | — | — | Egocentric cameras viewing gloved hands | not stated | No | open | — |
| CAIP pretraining video | 2026 | human-video | — | — | egocentric human video with extracted 3D hand keypoints | not stated | No | open | — |
| DYAD | 2026 | human-video | — | — | HoloLens 2 | not stated | No | open | — |
| Dyna-2 egocentric human video pre-training corpus | 2026 | human-video | — | — | head-mounted first-person cameras | not stated | No | open | — |
| EgoAVU | 2026 | human-video | — | — | existing egocentric videos | not stated | No | open | — |
| EgoCoT-Bench | 2026 | human-video | — | — | five public egocentric sets | not stated | No | open | — |
| EgoEVHands | 2026 | human-video | — | — | stereo event cameras, egocentric | not stated | No | open | — |
| EgoExoMem | 2026 | human-video | — | — | synchronised ego-exo videos | not stated | No | open | — |
| EgoFun3D | 2026 | human-video | — | — | egocentric videos with paired 3D geometry | not stated | No | open | — |
| EgoMAGIC | 2026 | human-video | — | — | head-mounted stereo camera with audio | CC BY 4.0 (per atlas) | Yes | open | — |
| EgoMemReason | 2026 | human-video | — | — | EgoLife videos | CC BY-NC 4.0 (per atlas) | No | open | — |
| EgoPlay | 2026 | human-video | — | — | mainly Ego4D clips | not stated | No | open | — |
| EgoProceVQA | 2026 | human-video | — | — | egocentric procedural videos | not stated | No | open | — |
| EgoScale | 2026 | human-video | — | — | egocentric human video | not stated | No | open | — |
| EgoSound | 2026 | human-video | — | — | Ego4D and EgoBlind videos | not stated | No | open | — |
| EgoSPT | 2026 | human-video | — | — | egocentric camera, fork pick-and-place | not stated | No | open | — |
| EgoStream | 2026 | human-video | — | — | several egocentric video sources | not stated | No | open | — |
| EgoTac corpus | 2026 | human-video | — | — | egocentric images with tactile labels | not stated | No | open | — |
| EgoTL | 2026 | human-video | — | — | Egocentric capture with think-aloud speech | not stated | No | open | — |
| EgoVid-Pro | 2026 | human-video | — | — | in-the-wild egocentric clips | not stated | No | open | — |
| EPIC-Contact | 2026 | human-video | — | — | EPIC-KITCHENS videos | CC BY-NC 4.0 (per atlas) | No | open | — |
| HA-2.2M | 2026 | human-video | — | — | Heterogeneous human videos, hand-centric | not stated | No | open | — |
| HandEdit | 2026 | human-video | — | — | Egocentric human hand clips mapped to 26 robot hand URDFs | not stated | No | open | — |
| HT-Bench | 2026 | human-video | — | — | egocentric camera with full-hand tactile sensing | not stated | No | open | — |
| HumanNet | 2026 | human-video | — | — | mixed first-person and third-person human video | not stated | No | open | — |
| Minerva-Ego | 2026 | human-video | — | — | HD-EPIC videos | not stated | No | open | — |
| MyEgo | 2026 | human-video | — | — | egocentric videos | not stated | No | open | — |
| Online TTM Dataset | 2026 | human-video | — | — | Ego4D social clips | not stated | No | open | — |
| OVO-S-Bench | 2026 | human-video | — | — | nine egocentric video sources | not stated | No | open | — |
| RetailSMV | 2026 | human-video | — | — | synchronised staff-view ego and exo cameras | not stated | No | open | — |
| SAW-Bench | 2026 | human-video | — | — | Ray-Ban Meta (Gen 2) smart glasses | CC BY-NC 4.0 (per atlas) | No | open | — |
| SG-Ego | 2026 | human-video | — | — | Ego4D videos | not stated | No | open | — |
| SlugTrails | 2026 | human-video | — | — | Aria glasses | not stated | No | open | — |
| TouchMoment | 2026 | human-video | — | — | egocentric videos | not stated | No | open | — |
| Universal Navigation Interface (UNI) | 2026 | human-video | — | — | four-wheeled rollator walker with smartphone, pushed by a person | not stated | No | open | — |
| WT++ | 2026 | human-video | — | — | handheld or body camera on urban walking tours | not stated | No | open | — |
| DGS-Net multi-object dexterous grasp dataset | 2026 | research-lab | — | — | dexterous hand in multi-object scenes | not stated | No | open | — |
| GCA-Bench | 2026 | research-lab | — | — | not stated in abstract | not stated | No | open | — |
| GraspIT | 2026 | research-lab | — | — | Franka Panda in NVIDIA Isaac Sim tabletop scenes | not stated | No | open | — |
| EgoSuite-Open100K | 2026 | human-video | — | 10,070 | human | custom | No | open | — |
| ArmnetBench v0.1 Dataset | 2026 | real-robot | — | 26 | bimanual_so_101, so_101 | Apache-2.0 | Yes | open | — |
| HABIT (Human Aware Behavior and Interaction Training dataset) | 2026 | real-robot | — | 164 | franka_panda | CC-BY-4.0 | Yes | open | — |
| HIW-500: Humanoids In-the-Wild | 2026 | human-video | — | 500 | unitree_humanoid | unknown | No | open | — |