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 |
|---|---|---|---|---|---|---|---|---|---|
| MARS-LVIG | 2024 | research-lab | — | — | UAV with downward LiDAR, camera, IMU, GNSS | not stated | No | open | — |
| QDGset | 2024 | research-lab | — | — | 6-DoF grasps, simulated | not stated | No | open | — |
| Sim-Grasp-Dataset | 2024 | research-lab | — | — | synthetic clutter, two-finger gripper | not stated | No | open | — |
| TARGO | 2024 | research-lab | — | — | synthetic plus some real cluttered scenes | not stated | No | open | — |
| ManiWAV bagel-in-the-wild | 2024 | research-lab | — | — | UMI gripper with contact microphone | not stated | No | open | — |
| Robot Utility Models (RUM) data | 2024 | research-lab | — | — | Stick V2 with iPhone; deployed on Hello Robot Stretch and xArm 7 | not stated for data (code MIT) | No | open | — |
| UMI on Legs manipulation trajectories | 2024 | research-lab | — | — | UMI handheld gripper (deployed on a quadruped with arm) | not stated | No | open | — |
| Nymeria and NymeriaPlus | 2024 | research-lab | — | — | not stated in fetched excerpt | separate dataset licence files; code CC BY-NC 4.0 | No | open | — |
| H2O / OmniH2O retargeted AMASS for H1 (amass_phc_filtered) | 2024 | research-lab | — | — | Unitree H1 | CC BY-NC 4.0 (repo) | No | open | — |
| HARPER | 2024 | research-lab | — | — | quadruped robot view of human | not stated | No | open | — |
| CORE4D | 2024 | research-lab | — | — | two humans jointly moving objects | not stated | No | open | — |
| FORCE | 2024 | research-lab | — | — | human plus object with force nuances | not stated | No | open | — |
| Harmony4D | 2024 | research-lab | — | — | closely interacting humans, multi-view video | not stated | No | open | — |
| LocoVR | 2024 | research-lab | — | — | two-person trajectories in VR homes | not stated | No | open | — |
| MotionLib (Being-M0) | 2024 | research-lab | — | — | human body motion with hierarchical text | not stated | No | open | — |
| MotionVerse (Large Motion Model) | 2024 | research-lab | — | — | human body, unified from 16 datasets | not stated | No | open | — |
| THOR-MAGNI | 2024 | research-lab | — | — | indoor human motion with robots | cc-by-4.0 | Yes | open | — |
| Pick-and-Place Human Multimodal Dataset | 2024 | research-lab | — | — | human pick and place | not stated | No | open | — |
| LLARVA visual trace pretraining data | 2024 | research-lab | — | — | Open X-Embodiment robots | not stated | No | open | — |
| TraceVLA visual trace prompting dataset | 2024 | research-lab | — | — | robot manipulation trajectories; evaluated on WidowX and SimplerEnv | not stated | No | open | — |
| NaVILA dataset | 2024 | research-lab | — | — | legged robots (VLA with locomotion skills) | not stated | No | open | — |
| RT-H language motions | 2024 | research-lab | — | — | not stated on project page | not stated | No | open | — |
| Can-Do | 2024 | research-lab | — | — | embodied planning samples | not stated | No | open | — |
| IDKB | 2024 | research-lab | — | — | driving handbooks and test questions | not stated | No | open | — |
| NatSGD and NatSGLD | 2024 | research-lab | — | — | speech, gesture and demonstrations for robot | not stated | No | open | — |
| ManipVQA | 2024 | research-lab | — | — | images with affordance and physical grounding QA | not stated | No | open | — |
| MSQA (Multi-modal Situated QA) | 2024 | research-lab | — | — | real-world 3D scenes | not stated | No | open | — |
| PhysiCLeAR (Octopi) | 2024 | research-lab | — | — | GelSight tactile videos with property annotations | not stated | No | open | — |
| RoboSpatial | 2024 | research-lab | — | — | real indoor and tabletop 3D scans with egocentric images | not stated | No | open | — |
| SpatialVLM spatial VQA data | 2024 | research-lab | — | — | internet real-world images | not stated | No | open | — |
| TLV (Touch-Language-Vision) | 2024 | research-lab | — | — | tactile images with sentence-level descriptions | not stated | No | open | — |
| Touch100k | 2024 | research-lab | — | — | GelSight tactile images with paired vision and text | not stated | No | open | — |
| TVL (Touch-Vision-Language) dataset | 2024 | research-lab | — | — | handheld 3D printed device with DIGIT sensor and Logitech BRIO webcam | not stated | No | open | — |
| VLA-3D | 2024 | research-lab | — | — | 3D scenes with referential language | not stated | No | open | — |
| AHA dataset (FailGen) | 2024 | research-lab | — | — | RLBench simulator | not stated | No | open | — |
| GOAT-Bench | 2024 | research-lab | — | — | Habitat scenes | not stated | No | open | — |
| HM-EQA | 2024 | research-lab | — | — | HM3D scenes | not stated | No | open | — |
| HM3D-OVON | 2024 | research-lab | — | — | HM3DSem scenes | MIT | Yes | open | — |
| In-building embodied delivery dataset | 2024 | research-lab | — | — | simulated building | not stated | No | open | — |
| LHPR-VLN | 2024 | research-lab | — | — | simulated indoor scenes | MIT | Yes | open | — |
| NaVid training data | 2024 | research-lab | — | — | Habitat continuous environments | not stated | No | open | — |
| NoisyEQA | 2024 | research-lab | — | — | EQA environments with noisy queries | not stated | No | open | — |
| OpenEQA | 2024 | research-lab | — | — | HM3D and real home and office scans | not stated | No | open | — |
| OpenUAV / TravelUAV | 2024 | research-lab | — | — | UAV simulation platform | not stated | No | open | — |
| S-EQA | 2024 | research-lab | — | — | VirtualHome simulator | not stated | No | open | — |
| SafeAgentBench | 2024 | research-lab | — | — | embodied task planning | not stated | No | open | — |
| SafetyDetect | 2024 | research-lab | — | — | home scenes | not stated | No | open | — |
| SAT (Spatial Aptitude Training) | 2024 | research-lab | — | — | ProcTHOR simulated indoor scenes | mit (Molmo2-ER mirror tag) | Yes | open | — |
| Uni-NaVid training data | 2024 | research-lab | — | — | simulated navigation across four sub-tasks | not stated | No | open | — |
| DROID language annotations and camera calibration (KarlP/droid) | 2024 | research-lab | — | — | DROID Franka platform | cc-by-4.0 (Hub tag) | Yes | open | — |
| RH20T-P | 2024 | research-lab | — | — | RH20T real-world robot data | not stated | No | open | — |
| AutoTrust | 2024 | research-lab | — | — | images from public driving sets | not stated | No | open | — |
| CODA-LM | 2024 | research-lab | — | — | CODA corner-case images | not stated | No | open | — |
| doScenes | 2024 | research-lab | — | — | nuScenes | not stated | No | open | — |
| MAPLM | 2024 | research-lab | — | — | HD map, lidar BEV, panoramic images | not stated | No | open | — |
| MM-AU | 2024 | research-lab | — | — | ego-view accident videos | not stated | No | open | — |
| NuInstruct | 2024 | research-lab | — | — | nuScenes multi-view video | not stated | No | open | — |
| OmniDrive | 2024 | research-lab | — | — | nuScenes | not stated | No | open | — |
| SURDS | 2024 | research-lab | — | — | nuScenes | not stated | No | open | — |
| WOMD-Reasoning | 2024 | research-lab | — | — | Waymo Open Motion scenarios | not stated | No | gated | — |
| CEAR | 2024 | research-lab | — | — | MIT Mini Cheetah with event camera, RGB-D, IMU, LiDAR, joint encoders | not stated | No | open | — |
| Leg-KILO dataset | 2024 | research-lab | — | — | Unitree Go1 with LiDAR, IMU, joint encoders, contact sensors | not stated | No | open | — |
| PFERD (Poses for Equine Research Dataset) | 2024 | research-lab | — | — | 5 horses, 100+ skin markers, 10 cameras, hSMAL model | not stated | No | open | — |
| TAIL and TAIL-Plus | 2024 | research-lab | — | — | wheeled and quadruped robots on sand | not stated | No | open | — |
| CoPeD | 2024 | research-lab | — | — | multiple robots | not stated | No | open | — |
| GEODE | 2024 | research-lab | — | — | heterogeneous LiDARs on multiple platforms | not stated | No | open | — |
| MCD (Multi-Campus Dataset) | 2024 | research-lab | — | — | ATV and handheld rigs with multiple lidars | not stated | No | open | — |
| SNAIL Radar | 2024 | research-lab | — | — | 4D radar rig on several platforms | not stated | No | open | — |
| VBR (Vision Benchmark in Rome) | 2024 | research-lab | — | — | handheld and vehicle rigs with RGB, LiDAR, IMU, GPS | not stated | No | open | — |
| DIDLM | 2024 | research-lab | — | — | infrared, depth cameras, LiDAR, 4D radar | not stated | No | open | — |
| ENWIDE LiDAR Inertial Dataset | 2024 | research-lab | — | — | LiDAR and IMU | not stated | No | open | — |
| Helvipad | 2024 | research-lab | — | — | omnidirectional stereo rig | not stated | No | open | — |
| Oxford Spires Dataset | 2024 | research-lab | — | — | perception unit with three cameras, LiDAR, IMU | not stated | No | open | — |
| Autonomous wheelchair road-crossing dataset | 2024 | research-lab | — | — | autonomous wheelchair | not stated | No | open | — |
| CSE Dataset (Collaborative SLAM in Service Environments) | 2024 | research-lab | — | — | multi-robot service environments | not stated | No | open | — |
| Earth Rover dataset recorded at the ICRA@40 party | 2024 | research-lab | — | — | FrodoBots Earth Rover | not stated | No | open | — |
| RoboSense | 2024 | research-lab | — | — | mobile robot with cameras, LiDAR and fisheye | not stated | No | open | — |
| Robot@Home2 | 2024 | research-lab | — | — | mobile robot in home environments | cc-by-4.0 | Yes | open | — |
| Semantic2D | 2024 | research-lab | — | — | 2D lidar on mobile robot | not stated | No | open | — |
| THUD++ | 2024 | research-lab | — | — | real robot platform plus physical simulation | not stated | No | open | — |
| TiROD | 2024 | research-lab | — | — | tiny mobile robot camera | not stated | No | open | — |
| UMAD | 2024 | research-lab | — | — | mobile robot patrol images | not stated | No | open | — |
| BridgeData V2 | 2023 | real-robot | — | 111.1 | WidowX 250s | MIT | Yes | open | hdf5, zarr, mp4 |
| ALOHA — Action Chunking with Transformers | 2023 | real-robot | — | — | ALOHA bimanual robot, ViperX 300s, WidowX 250s | MIT | Yes | open | hdf5, lerobot |
| RoboSet — A Diverse Multi-Task Robot Learning Dataset | 2023 | real-robot | — | — | Franka Panda | CC BY 4.0 | Yes | open | hdf5, lerobot |
| Open X-Embodiment — Robotic Learning Datasets | 2023 | real-robot | — | — | Franka Panda, Google RT-2, WidowX 250s | Apache 2.0 | Yes | open | zarr, tfrecord |
| PushT — Diffusion Policy Reference Dataset | 2023 | synthetic | — | — | simulated 2D pusher | Apache 2.0 | Yes | open | lerobot, parquet |
| RflyMAD | 2023 | research-lab | — | — | multicopters, simulated and real | not stated | No | open | — |
| WiSARD | 2023 | research-lab | — | — | UAV visual and thermal | not stated | No | open | — |
| GRACO | 2023 | research-lab | — | — | UGV and UAV | not stated | No | open | — |
| I-24 MOTION | 2023 | research-lab | — | — | 276 pole-mounted cameras over 4.2 miles of freeway | not stated | No | gated | — |
| ScenarioNet | 2023 | research-lab | — | — | Waymo, nuScenes, nuPlan scenarios in MetaDrive | not stated | No | open | — |
| AIDE | 2023 | research-lab | — | — | multi-view in-cabin and out-of-cabin cameras | not stated | No | open | — |
| Aurora Multi-Sensor Dataset | 2023 | research-lab | — | — | Uber ATG vehicles, lidar and cameras, Pittsburgh | not stated | No | open | — |
| AVOID | 2023 | research-lab | — | — | AV incident reports | not stated | No | open | — |
| Dual Radar | 2023 | research-lab | — | — | car with two 4D radars, lidar, camera | not stated | No | open | — |
| ECMD | 2023 | research-lab | — | — | event cameras with multi-sensor suite | not stated | No | open | — |
| HeLiPR | 2023 | research-lab | — | — | car with heterogeneous lidars | not stated | No | open | — |
| IDD-AW | 2023 | research-lab | — | — | RGB and NIR camera pairs | not stated | No | gated | — |
| Occ3D | 2023 | research-lab | — | — | Waymo and nuScenes | not stated | No | open | — |