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 |
|---|---|---|---|---|---|---|---|---|---|
| Boreas | 2022 | research-lab | — | — | car with Velodyne Alpha Prime, Navtech radar, FLIR camera | not stated | No | open | — |
| CODA | 2022 | research-lab | — | — | images from KITTI, nuScenes, ONCE | CC BY-NC-SA 4.0 | No | open | — |
| DAIR-V2X | 2022 | research-lab | — | — | vehicle and roadside cameras and lidar | not stated | No | gated | — |
| DualCam | 2022 | research-lab | — | — | dual camera traffic lights | not stated | No | open | — |
| IDD-3D | 2022 | research-lab | — | — | car with cameras and lidar | not stated | No | gated | — |
| Ithaca365 | 2022 | research-lab | — | — | car with cameras and lidar | CC BY-NC-SA 4.0 | No | open | — |
| K-Lane | 2022 | research-lab | — | — | lidar and camera | Apache-2.0 | Yes | open | — |
| K-Radar | 2022 | research-lab | — | — | car with 4D radar, lidar, stereo cameras | CC BY-NC-ND | No | open | — |
| ONCE-3DLanes | 2022 | research-lab | — | — | ONCE vehicle | not stated | No | gated | — |
| OpenLane | 2022 | research-lab | — | — | Waymo Open images with 3D lanes | not stated | No | open | — |
| PSI | 2022 | research-lab | — | — | dashcam video | not stated | No | open | — |
| SceNDD | 2022 | research-lab | — | — | naturalistic driving | not stated | No | open | — |
| TJ4DRadSet | 2022 | research-lab | — | — | car with 4D radar, lidar, camera | not stated | No | open | — |
| View-of-Delft | 2022 | research-lab | — | — | car with 64-layer lidar, stereo camera, 3+1D radar | not stated | No | open | — |
| WildDash 2 | 2022 | research-lab | — | — | dashcam images worldwide | registration required | No | gated | — |
| ActionSense | 2022 | human-video | — | — | eye tracker with first-person camera, EMG, 17-IMU suit, tactile gloves | not stated | No | open | — |
| ARCTIC | 2022 | human-video | — | — | mocap studio with static cameras and an egocentric camera | not stated | No | open | — |
| AssistQ | 2022 | human-video | — | — | newly filmed egocentric instructional videos | not stated | No | open | — |
| EDINA | 2022 | human-video | — | — | Egocentric RGB with depth and surface normals | not stated | No | open | — |
| EgoHOS | 2022 | human-video | — | — | frames from several egocentric sets | MIT (per atlas) | Yes | open | — |
| EgoTaskQA | 2022 | human-video | — | — | LEMMA egocentric videos | not stated | No | open | — |
| FT-HID | 2022 | human-video | — | — | RGB-D first- and third-person capture | not stated | No | open | — |
| GIMO | 2022 | human-video | — | — | head-mounted device with gaze, inertial motion capture | not stated | No | open | — |
| HOI4D | 2022 | human-video | — | — | head-mounted RGB-D | CC BY-NC 4.0 (per atlas) | No | open | — |
| SceneEgo (EgoGTA, EgoPW-Scene) | 2022 | human-video | — | — | head-mounted fisheye camera, synthetic and real | not stated | No | open | — |
| UmeTrack | 2022 | human-video | — | — | VR headset with 4 fisheye sensors | CC BY-NC 4.0 | No | open | — |
| VOST | 2022 | human-video | — | — | egocentric videos from existing sets | not stated | No | open | — |
| OakInk | 2022 | human-video | — | — | multi-camera third-person rig (not egocentric) | CC-BY-NC-SA-3.0 (Hub oakink/OakInk-v1) | No | open | — |
| UnrealEgo / UnrealEgo2 / UnrealEgo-RW | 2022 | human-video | — | — | virtual eyeglasses with two fisheye cameras; RW adds real device | not stated | No | open | — |
| DA2 | 2022 | research-lab | — | — | dual-arm parallel-jaw grasp pairs | not stated | No | open | — |
| DexGraspNet | 2022 | research-lab | — | — | ShadowHand, validated in Isaac Gym | CC BY-NC 4.0 | No | open | — |
| MultiDex (GenDexGrasp) | 2022 | research-lab | — | — | multiple robotic hands | not stated | No | open | — |
| DREDS and STD | 2022 | research-lab | — | — | simulated and real RGB-D of specular, transparent, diffuse objects | CC BY-NC 4.0 | No | open | — |
| TransCG | 2022 | research-lab | — | — | RGB-D of transparent objects, robot-collected | CC BY-NC-SA 4.0 | No | open | — |
| RT-1 Robot Action Dataset | 2022 | human-video | — | 350.6 | — | CC-BY-4.0 | Yes | open | — |
| CALVIN | 2022 | synthetic | — | 24 | franka_panda | MIT | Yes | open | — |
| Language-Table | 2022 | real-robot | — | 3,865 | xarm | Apache-2.0 | Yes | open | — |
| Ego4D | 2022 | human-video | — | 3,670 | human | research-only | No | open | — |
| EPIC-KITCHENS-100 | 2022 | human-video | — | 100 | human | CC-BY-NC-4.0 | No | open | — |
| TartanDrive | 2022 | human-video | — | 5 | offroad_vehicle | CC-BY-4.0 | Yes | open | — |
| ALTO | 2022 | research-lab | — | — | helicopter with downward camera, GPS-INS, laser altimeter | not stated | No | open | — |
| DenseUAV | 2022 | research-lab | — | — | low-altitude UAV plus satellite | not stated | No | open | — |
| HIT-UAV | 2022 | research-lab | — | — | UAV infrared thermal | not stated | No | open | — |
| INSANE | 2022 | research-lab | — | — | multi-sensor MAV | not stated | No | open | — |
| NTU VIRAL | 2022 | research-lab | — | — | UAV with two 3D lidars, two cameras, IMUs, UWB | not stated | No | open | — |
| Rope3D | 2022 | research-lab | — | — | roadside cameras with lidar labels | registration required (per ad-datasets) | No | gated | — |
| MetaGraspNet / MetaGraspNetV2 | 2022 | research-lab | — | — | synthetic and real RGBD bin scenes, parallel-jaw and vacuum labels | see repo License.txt | No | open | — |
| PoseIt | 2022 | research-lab | — | — | robot arm grasping with GelSight and force sensing, RGB-D | not stated | No | open | — |
| BEHAVE | 2022 | research-lab | — | — | human body plus object, multi-view RGBD, SMPL fits | not stated | No | open | — |
| COUCH | 2022 | research-lab | — | — | human body with chairs | not stated | No | open | — |
| FLAG3D | 2022 | research-lab | — | — | human fitness activity, mocap plus rendered and real video | not stated | No | open | — |
| HODome (NeuralDome) | 2022 | research-lab | — | — | human plus object in dense multi-view dome | not stated | No | open | — |
| HuMMan | 2022 | research-lab | — | — | human body, RGB-D, point clouds, SMPL, textured meshes | not stated | No | open | — |
| InterCap | 2022 | research-lab | — | — | human plus object, 6 RGB-D cameras | not stated | No | open | — |
| RICH | 2022 | research-lab | — | — | human bodies with scene scans, multi-view 4K video | not stated | No | open | — |
| ZeroEGGS | 2022 | research-lab | — | — | human full body gesture with fingers and speech | not stated | No | open | — |
| ASE sword and shield motions | 2022 | research-lab | — | — | simulated humanoid character | not stated | No | open | — |
| CHICO | 2022 | research-lab | — | — | human and cobot in industrial tasks | not stated | No | open | — |
| SQA3D | 2022 | research-lab | — | — | 650 ScanNet scenes | CC-BY-4.0 (data), Apache (code) | No | open | — |
| DialFRED | 2022 | research-lab | — | — | AI2-THOR | not stated | No | open | — |
| Housekeep | 2022 | research-lab | — | — | Habitat households | not stated | No | open | — |
| IVLN (Iterative VLN) | 2022 | research-lab | — | — | Matterport3D and Habitat | not stated | No | open | — |
| DRAMA | 2022 | research-lab | — | — | vehicle camera, Tokyo | research only | No | open | — |
| AMP for hardware A1 mocap motions | 2022 | research-lab | — | — | Unitree A1 | not stated | No | open | — |
| Cerberus visual-inertial-leg odometry data | 2022 | research-lab | — | — | Unitree quadrupeds | not stated | No | open | — |
| UTIL | 2022 | research-lab | — | — | Decawave DWM1000 UWB modules | not stated | No | open | — |
| FusionPortable | 2022 | research-lab | — | — | portable multi-sensor suite on diverse platforms | not stated | No | open | — |
| OdomBeyondVision | 2022 | research-lab | — | — | UAV, UGV and handheld with radar, thermal and other sensors | not stated | No | open | — |
| EVIMO2 | 2022 | research-lab | — | — | event cameras | not stated | No | open | — |
| GSV-Cities | 2022 | research-lab | — | — | Google Street View place images | not stated | No | open | — |
| Hilti-Oxford Dataset (2022) | 2022 | research-lab | — | — | handheld LiDAR, camera, IMU rig | not stated | No | open | — |
| Multi-Modal Lidar Dataset (general-purpose localization and mapping) | 2022 | research-lab | — | — | multiple lidar types | not stated | No | open | — |
| T2FPV | 2022 | research-lab | — | — | first-person pedestrian trajectories | not stated | No | open | — |
| VECtor | 2022 | research-lab | — | — | event-centric multi-sensor rig | not stated | No | open | — |
| ViViD++ | 2022 | research-lab | — | — | thermal, event, RGB-D, lidar sensors | not stated | No | open | — |
| ALITA | 2022 | research-lab | — | — | LiDAR and omnidirectional camera | not stated | No | open | — |
| S3E | 2022 | research-lab | — | — | fleet of unmanned ground vehicles | not stated | No | open | — |
| TGRMPT | 2022 | research-lab | — | — | tour-guide robot | not stated | No | open | — |
| TorWIC (Mapping and SLAM) | 2022 | research-lab | — | — | mobile robot in Clearpath Robotics warehouse | not stated | No | open | — |
| RoboMimic — A Study of Imitation Learning from Observations | 2021 | real-robot | — | — | Franka Panda, Sawyer | MIT | Yes | open | hdf5 |
| SeaDronesSee | 2021 | research-lab | — | — | UAV over open water | not stated | No | open | — |
| VID (Visual-Inertial-Dynamical) Dataset | 2021 | research-lab | — | — | multirotor with rotor speed and force sensing | not stated | No | open | — |
| Automatum Open Highway | 2021 | research-lab | — | — | drone over German highways | CC BY-ND | No | open | — |
| Topo-boundary | 2021 | research-lab | — | — | aerial images of road boundaries | not stated | No | open | — |
| ACDC | 2021 | research-lab | — | — | car camera, Switzerland | not stated | No | gated | — |
| DSEC | 2021 | research-lab | — | — | stereo event cameras, RGB, lidar, GPS | not stated | No | open | — |
| DurLAR | 2021 | research-lab | — | — | car with 128-channel lidar, stereo camera, lux meter, GNSS/INS | not stated | No | open | — |
| Honda LOKI | 2021 | research-lab | — | — | vehicle camera and lidar | research only | No | open | — |
| KITTI-360 | 2021 | research-lab | — | — | car with fisheye and perspective cameras, lidar | CC BY-NC-SA 3.0 | No | open | — |
| Oxford Road Boundaries | 2021 | research-lab | — | — | Oxford RobotCar | not stated | No | open | — |
| PixSet | 2021 | research-lab | — | — | car with full-waveform flash lidar, cameras, radar | not stated | No | open | — |
| PVDN (Provident Vehicle Detection at Night) | 2021 | research-lab | — | — | grayscale camera at night | not stated | No | open | — |
| RADIal | 2021 | research-lab | — | — | car with HD radar, camera, laser scanner | not stated | No | open | — |
| ROAD | 2021 | research-lab | — | — | Oxford RobotCar videos | CC BY-SA 4.0 | Yes | open | — |
| SegmentMeIfYouCan (RoadAnomaly21, RoadObstacle21) | 2021 | research-lab | — | — | road images with anomalies | various | No | open | — |
| Shifts (vehicle motion prediction) | 2021 | research-lab | — | — | Yandex self-driving fleet tracks | not stated | No | open | — |
| SODA10M | 2021 | research-lab | — | — | dashcam images | CC BY-NC-SA 4.0 | No | gated | — |
| H2O | 2021 | human-video | — | — | synchronised multi-view RGB-D including egocentric | not stated | No | open | — |
| HPS (Human POSEitioning System) | 2021 | research-lab | — | — | human body, body-worn IMUs plus head camera in scanned scenes | not stated | No | open | — |
| N-EPIC-Kitchens | 2021 | human-video | — | — | EPIC-KITCHENS videos converted to event streams | not stated | No | open | — |