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 |
|---|---|---|---|---|---|---|---|---|---|
| DADA-2000 | 2019 | research-lab | — | — | dashcam accident videos with driver attention | not stated | No | open | — |
| DrivingStereo | 2019 | research-lab | — | — | car with stereo cameras and lidar | not stated | No | open | — |
| EU Long-term Dataset | 2019 | research-lab | — | — | multi-sensor vehicle | not stated | No | open | — |
| Fishyscapes | 2019 | research-lab | — | — | Cityscapes and LostAndFound images with anomalies | not stated | No | open | — |
| Honda H3D | 2019 | research-lab | — | — | HDD vehicle, 360 lidar | research only | No | open | — |
| Honda HEV-I | 2019 | research-lab | — | — | egocentric vehicle camera at intersections | research only | No | open | — |
| MulRan | 2019 | research-lab | — | — | car with radar and lidar | registration required | No | open | — |
| nuScenes | 2019 | research-lab | — | — | car with 6 cameras, lidar, 5 radars | CC BY-NC-SA 4.0 | No | gated | — |
| nuScenes, nuImages, nuPlan family | 2019 | research-lab | — | — | Motional vehicles | not stated in fetched excerpt | No | open | — |
| SeeingThroughFog (DENSE) | 2019 | research-lab | — | — | car with gated camera, radar, lidar, FIR | research and teaching | No | gated | — |
| SemanticKITTI | 2019 | research-lab | — | — | KITTI odometry lidar sequences | CC BY-NC-SA 4.0 | No | open | — |
| Unsupervised LLAMAS | 2019 | research-lab | — | — | highway camera with lidar-map lane labels | non-commercial research | No | gated | — |
| UrbanLoco | 2019 | research-lab | — | — | car with full sensor suite, San Francisco and Hong Kong | not stated | No | open | — |
| Waymo Open Dataset | 2019 | research-lab | — | — | Waymo vehicles | Waymo terms (sign-in) | No | open | — |
| Waymo Open Dataset Perception | 2019 | research-lab | — | — | Waymo vehicle, 5 lidars and 5 cameras | Waymo non-commercial licence | No | gated | — |
| WoodScape | 2019 | research-lab | — | — | 4 fisheye surround cameras, lidar | non-commercial research | No | gated | — |
| TEgO | 2019 | human-video | — | — | smartphone camera | not stated | No | open | — |
| You2Me | 2019 | human-video | — | — | chest-mounted camera (per awesome list) | not stated | No | open | — |
| HO-3D | 2019 | human-video | — | — | one or several fixed RGB-D cameras (not egocentric) | not stated | No | open | — |
| ContactDB | 2019 | research-lab | — | — | human hand contact maps on objects | not stated | No | open | — |
| ClearGrasp | 2019 | research-lab | — | — | RGB-D of transparent objects | not stated | No | open | — |
| 6-DOF GraspNet data | 2019 | research-lab | — | — | simulated parallel-jaw grasps | see repo LICENSE | No | open | — |
| A Billion Ways to Grasp | 2019 | research-lab | — | — | simulated parallel-jaw grasps, dense SE(3) sampling | not stated | No | open | — |
| RoboNet | 2019 | real-robot | — | — | fetch, franka_panda, kuka_iiwa | CC-BY-4.0 | Yes | open | — |
| ALFA | 2019 | research-lab | — | 1.3 | fixed-wing UAV | CC-BY-4.0 | No | open | — |
| Tsinghua Visual-Tactile Dataset (Multimodal Grasp Data Set) | 2019 | research-lab | — | — | not stated | not stated | No | open | — |
| AI4Animation human mocap sets (NSM 2019, Local Motion Phases 2020, Cranberry 2024) | 2019 | research-lab | — | — | human body mocap | CC BY-NC 4.0 | No | open | — |
| NTU RGB+D 120 | 2019 | research-lab | — | — | human body, Kinect RGB-D skeletons | not stated | No | open | — |
| PROX | 2019 | research-lab | — | — | human body in scanned 3D scenes, RGB-D | not stated | No | open | — |
| FreiHAND | 2019 | research-lab | — | — | RGB hand pose and shape | not stated | No | open | — |
| ScanRefer | 2019 | research-lab | — | — | 800 ScanNet scenes | not stated | No | open | — |
| CVDN (Vision-and-Dialog Navigation) | 2019 | research-lab | — | — | Matterport3D simulator | not stated | No | open | — |
| MP3D-EQA | 2019 | research-lab | — | — | Matterport3D | not stated | No | open | — |
| MT-EQA | 2019 | research-lab | — | — | simulated houses, multi-target questions | not stated | No | open | — |
| R4R (Room-for-Room) | 2019 | research-lab | — | — | Matterport3D simulator | not stated | No | open | — |
| REVERIE | 2019 | research-lab | — | — | Matterport3D simulator | not stated | No | open | — |
| Talk2Nav | 2019 | research-lab | — | — | Street View | not stated | No | open | — |
| Talk2Car | 2019 | research-lab | — | — | nuScenes frames | MIT | Yes | open | — |
| JRDB (JackRabbot Dataset and Benchmark) | 2019 | research-lab | — | — | JackRabbot social mobile manipulator | CC BY-NC-SA 3.0 | No | gated | — |
| THOR | 2019 | research-lab | — | — | motion capture room with a mobile robot and 3D lidar | not stated | No | open | — |
| ETH3D SLAM benchmark | 2019 | research-lab | — | — | not stated | not stated | No | open | — |
| Oxford Multimotion Dataset | 2019 | research-lab | — | — | static and dynamic sensors | not stated | No | open | — |
| OpenLORIS-Scene | 2019 | research-lab | — | — | service robots | not stated | No | open | — |
| UAVid | 2018 | research-lab | — | — | UAV 4K slanted-view video | not stated | No | open | — |
| VisDrone | 2018 | research-lab | — | — | drone cameras | not stated | No | open | — |
| Audi fleet CAN dataset (Drive2Vec) | 2018 | research-lab | — | — | Audi test vehicles with data loggers | not stated | No | open | — |
| CULane | 2018 | research-lab | — | — | cameras on six vehicles, Beijing | non-commercial | No | open | — |
| EuroCity Persons | 2018 | research-lab | — | — | car camera, 31 European cities | research use | No | open | — |
| India Driving Dataset (IDD) | 2018 | research-lab | — | — | car front camera, Hyderabad and Bangalore | not stated | No | gated | — |
| N-CARS | 2018 | research-lab | — | — | event camera car classification | not stated | No | open | — |
| PedX | 2018 | research-lab | — | — | stereo cameras and LiDAR at intersections | not stated | No | open | — |
| EgoYouTubeHands / HandOverFace | 2018 | human-video | — | — | YouTube egocentric videos | not stated | No | open | — |
| Mo2Cap2 | 2018 | human-video | — | — | cap-mounted fisheye camera | not stated | No | open | — |
| Toybox | 2018 | human-video | — | — | Egocentric video of handheld object transformations | not stated | No | open | — |
| 40K naturalistic 6-DoF grasp demonstrations | 2018 | research-lab | — | — | RightHand Robotics three-fingered ReFlex gripper | not stated | No | open | — |
| Robot Learning in Homes grasping dataset | 2018 | research-lab | — | — | low cost mobile robot arm in homes | not stated | No | open | — |
| VMRD (Visual Manipulation Relationship Dataset) | 2018 | research-lab | — | — | RGB images of stacked objects | not stated | No | open | — |
| Yamaha-CMU Off-Road Dataset (YCOR) | 2018 | human-video | — | — | — | CC-BY-4.0 | Yes | open | — |
| Jacquard | 2018 | research-lab | — | — | synthetic RGB-D with parallel-jaw grasp labels | signed EULA | No | open | — |
| More Than a Feeling | 2018 | research-lab | — | — | two-finger gripper with GelSight sensors on each finger | not stated | No | open | — |
| Slip Detection with Combined Tactile and Visual Information | 2018 | research-lab | — | — | robot arm gripper with GelSight sensor and camera | not stated | No | open | — |
| 3DPW (3D Poses in the Wild) | 2018 | research-lab | — | — | human body, IMUs plus moving phone camera | not stated | No | open | — |
| DeepMimic and AMP reference motion clips | 2018 | research-lab | — | — | simulated humanoid characters | not stated | No | open | — |
| LANI and CHAI | 2018 | research-lab | — | — | simulated 3D environments | not stated | No | open | — |
| Talk the Walk | 2018 | research-lab | — | — | New York City 360 imagery | not stated | No | open | — |
| Touchdown | 2018 | research-lab | — | — | Street View panoramas | not stated | No | open | — |
| VirtualHome (ActivityPrograms) | 2018 | research-lab | — | — | Unity3D household simulator | not stated | No | open | — |
| BDD-X | 2018 | research-lab | — | — | BDD dashcam video | not stated | No | open | — |
| AI4Animation quadruped mocap (Mode-Adaptive Neural Networks) | 2018 | research-lab | — | — | dog mocap | CC BY-NC 4.0 | No | open | — |
| MVSEC | 2018 | research-lab | — | — | stereo event cameras on car, motorbike, hexacopter, handheld | not stated | No | open | — |
| ADVIO | 2018 | research-lab | — | — | iPhone, Tango, Pixel | not stated | No | open | — |
| OxIOD | 2018 | research-lab | — | — | phone IMUs | not stated | No | open | — |
| TUM VI Benchmark | 2018 | research-lab | — | — | handheld stereo camera and IMU | not stated | No | open | — |
| AdobeIndoorNav | 2018 | research-lab | — | — | real indoor robot navigation | not stated | No | open | — |
| Zurich Urban MAV Dataset | 2017 | research-lab | — | — | Fotokite MAV with GoPro, GPS, IMU | no restriction | No | open | — |
| Mapillary Vistas | 2017 | research-lab | — | — | crowdsourced street-level cameras | CC BY-NC-SA (per ad-datasets) | No | gated | — |
| CityPersons | 2017 | research-lab | — | — | Cityscapes pedestrian boxes | not stated | No | open | — |
| Complex Urban (KAIST Urban) | 2017 | research-lab | — | — | car with lidars, stereo camera, GPS/IMU, altimeter | CC BY-NC-SA 4.0 | No | open | — |
| DDD17 | 2017 | research-lab | — | — | DAVIS event camera on car | not stated | No | open | — |
| NEXET | 2017 | research-lab | — | — | Nexar dashcam images | not stated | No | open | — |
| Paris-Lille-3D | 2017 | research-lab | — | — | mobile LiDAR urban point cloud | not stated | No | open | — |
| UDRIVE naturalistic driving database | 2017 | research-lab | — | — | instrumented cars, trucks and powered two-wheelers | not stated | No | open | — |
| EgoDexter | 2017 | human-video | — | — | egocentric RGB-D sensor | not stated | No | open | — |
| Something-Something | 2017 | human-video | — | — | Crowd workers filming their own hands | not stated | No | open | — |
| THU-READ | 2017 | human-video | — | — | RGB-D sensor mounted on a helmet | not stated | No | open | — |
| Dex-Net 2.0 | 2017 | research-lab | — | — | synthetic point clouds, parallel-jaw grasps | not stated | No | open | — |
| Wire Detection Dataset | 2017 | human-video | — | — | drone | CC-BY-4.0 | Yes | open | — |
| Something-Something V2 | 2017 | research-lab | — | — | human hands with everyday objects | other | No | open | — |
| EgoGesture | 2017 | human-video | — | — | head-mounted RGB-D camera | not stated | No | open | — |
| Integrated grasping and vision simulator dataset (Veres) | 2017 | research-lab | — | — | simulated multi-fingered dexterous hand | not stated | No | open | — |
| MIT-Princeton ARC 2017 grasping and suction data | 2017 | research-lab | — | — | robot picking system, RGB-D | not stated | No | open | — |
| The Feeling of Success | 2017 | research-lab | — | — | two-finger gripper with GelSight sensors on each finger | not stated | No | open | — |
| TotalCapture | 2017 | research-lab | — | — | human body, multi-view video plus IMU plus Vicon | not stated | No | open | — |
| EQA (Embodied Question Answering) | 2017 | research-lab | — | — | simulated houses | not stated | No | open | — |
| IQUAD V1 | 2017 | research-lab | — | — | AI2-THOR | not stated | No | open | — |
| R2R (Room-to-Room) | 2017 | research-lab | — | — | Matterport3D simulator | not stated | No | open | — |
| Aachen Day-Night, RobotCar Seasons, CMU Seasons | 2017 | research-lab | — | — | outdoor visual localization | not stated | No | open | — |
| PennCOSYVIO | 2017 | research-lab | — | — | handheld visual-inertial rig | not stated | No | open | — |
| Active Vision Dataset | 2017 | research-lab | — | — | robot-captured dense RGB-D | not stated | No | open | — |
| L-CAS 3D Point Cloud People Dataset | 2017 | research-lab | — | — | 3D LiDAR on mobile robot | not stated | No | open | — |