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 |
|---|---|---|---|---|---|---|---|---|---|
| RoboFactory dataset | 2025 | synthetic | — | — | multi-robot ManiSkill scenes | mit | Yes | open | — |
| T2M4LVO | 2025 | research-lab | — | — | non-human skeletons from Truebones Zoo | CC BY-NC 4.0 | No | open | — |
| hqfang/memorybench | 2025 | research-lab | — | — | not stated | apache-2.0 | Yes | open | — |
| PHUMA | 2025 | research-lab | — | — | Unitree G1 configuration on the card | apache-2.0 | Yes | open | — |
| Arena-GR1-Manipulation-Task | 2025 | synthetic | — | — | Fourier GR1 in Isaac Lab Arena | cc-by-4.0 | Yes | open | — |
| AgriLiRa4D | 2025 | research-lab | — | — | multi-sensor UAV over farmland | not stated | No | open | — |
| AnyVisLoc | 2025 | research-lab | — | — | low-altitude UAV with 2.5D reference maps | not stated | No | open | — |
| LASED | 2025 | research-lab | — | — | aerial orthophotos | not stated | No | open | — |
| MATRIX multi-drone multi-view pedestrian dataset | 2025 | research-lab | — | — | multiple drones | not stated | No | open | — |
| MESSI | 2025 | research-lab | — | — | drone over dense urban area | not stated | No | open | — |
| MMLA | 2025 | research-lab | — | — | low-altitude drone wildlife video | not stated | No | open | — |
| OrthoLoC | 2025 | research-lab | — | — | UAV images with orthophoto and DSM references | not stated | No | open | — |
| EgoExoBench | 2025 | human-video | — | — | public ego-exo datasets | not stated | No | gated | — |
| GRIP | 2025 | research-lab | — | — | soft and rigid grippers, IPC simulation | not stated | No | open | — |
| Parcel-Suction-Dataset | 2025 | research-lab | — | — | synthetic cluttered parcel scenes, suction | not stated | No | open | — |
| PRISM (GraspMolmo) | 2025 | research-lab | — | — | synthetic cluttered scenes with task language and grasps | not stated | No | open | — |
| Supermarket-6DoF | 2025 | research-lab | — | — | physical robot grasp executions | not stated | No | open | — |
| ActiveUMI | 2025 | research-lab | — | — | VR controllers shaped as grippers plus head-mounted display | not stated | No | open | — |
| Generalist AI GEN-0 training corpus (November 2025 snapshot) | 2025 | research-lab | — | — | "1,000s of data collection devices and robots" (not specified further) | not stated | No | open | — |
| MV-UMI | 2025 | research-lab | — | — | UMI gripper with added third-person camera | not stated | No | open | — |
| RDT2 UMI dataset | 2025 | research-lab | — | — | redesigned bimanual UMI gripper with HTC Vive tracking | not stated | No | open | — |
| ASAP motions | 2025 | research-lab | — | — | Unitree G1 29 DoF annealed to 23 DoF | MIT (repo) | No | open | — |
| SONIC motion capture training set | 2025 | research-lab | — | — | human motion capture used for humanoid whole-body control | not stated | No | open | — |
| TWIST motion dataset | 2025 | research-lab | — | — | Unitree G1 | MIT (repo) | No | open | — |
| 3HANDS | 2025 | research-lab | — | — | handovers with supernumerary limb | not stated | No | open | — |
| ClimbingCap (AscendMotion) | 2025 | research-lab | — | — | human climbers, RGB, LiDAR, IMU | not stated | No | open | — |
| HHI-Assist | 2025 | research-lab | — | — | human-human physical assistance | not stated | No | open | — |
| HUMOTO | 2025 | research-lab | — | — | human body with articulated objects | not stated | No | open | — |
| SD-Dataset | 2025 | research-lab | — | — | human assembly and disassembly near robots | not stated | No | open | — |
| YCB-Handovers | 2025 | research-lab | — | — | human-human handovers, YCB objects | not stated | No | open | — |
| K2MUSE | 2025 | research-lab | — | — | lower-limb kinematics, kinetics, ultrasound, EMG | not stated | No | open | — |
| Waymo-3DSkelMo | 2025 | research-lab | — | — | Waymo lidar pedestrians | not stated | No | gated | — |
| AerialVG | 2025 | research-lab | — | — | real aerial images | not stated | No | gated | — |
| AirCopBench | 2025 | research-lab | — | — | multi-drone, simulator and real data | not stated | No | open | — |
| MM-UAVBench | 2025 | research-lab | — | — | real UAV data from public datasets | apache-2.0 | Yes | open | — |
| Open3D-VQA | 2025 | research-lab | — | — | open-space aerial views | not stated | No | open | — |
| UrbanVideo-Bench | 2025 | research-lab | — | — | manually controlled drones in real and simulated cities | not stated | No | open | — |
| GuardianFail-36k and UR5-Fail | 2025 | research-lab | — | — | RLBench simulator, BridgeDataV2 robot, UR5 with three cameras | not stated | No | open | — |
| RoVI-Book | 2025 | research-lab | — | — | robotic visual instruction images from OXE | not stated | No | open | — |
| MUSON | 2025 | research-lab | — | — | egocentric views in social scenes | not stated | No | open | — |
| SAIT guide dog robot dataset | 2025 | research-lab | — | — | guide dog robots | not stated | No | open | — |
| AmbiK | 2025 | research-lab | — | — | kitchen task text | not stated | No | open | — |
| CausalVQA | 2025 | research-lab | — | — | real-world video | not stated | No | open | — |
| J-ORA | 2025 | research-lab | — | — | robot perception, Japanese dialogue | not stated | No | open | — |
| MVPBench (Minimal Video Pairs) | 2025 | research-lab | — | — | real and simulated video pairs | not stated | No | open | — |
| RoboDesign1M | 2025 | research-lab | — | — | robot design documents | not stated | No | open | — |
| User Questions for Explainable Robotics | 2025 | research-lab | — | — | household robots | not stated | No | open | — |
| VeBrain-600k | 2025 | research-lab | — | — | not stated in abstract | not stated | No | open | — |
| WorldPrediction | 2025 | research-lab | — | — | procedural human activity video | not stated | No | open | — |
| A4Bench | 2025 | research-lab | — | — | images | not stated | No | open | — |
| CLTP tactile point cloud-language dataset | 2025 | research-lab | — | — | tactile 3D point clouds | not stated | No | open | — |
| ERQA | 2025 | research-lab | — | — | real-world robotics-focused images | CC-BY-4.0 | Yes | open | — |
| MoniRefer | 2025 | research-lab | — | — | roadside lidar and cameras | not stated | No | open | — |
| ObjTac | 2025 | research-lab | — | — | force-based tactile sensor (Paxini, per TaCo table) | not stated | No | open | — |
| RAGNet | 2025 | research-lab | — | — | images with affordance masks and instructions | not stated | No | open | — |
| RoboAfford++ | 2025 | research-lab | — | — | images for manipulation and navigation affordance | not stated | No | open | — |
| SURPRISE3D | 2025 | research-lab | — | — | 3D indoor scenes | not stated | No | open | — |
| TUMTraffic-VideoQA | 2025 | research-lab | — | — | roadside cameras | not stated | No | open | — |
| ViewSpatial-Bench | 2025 | research-lab | — | — | multi-perspective spatial QA | not stated | No | open | — |
| VTV150K | 2025 | research-lab | — | — | GelSight Mini, DIGIT, Tac3D | not stated | No | open | — |
| CityEQA-EC | 2025 | research-lab | — | — | 3D urban simulator | not stated | No | open | — |
| DriveQA | 2025 | research-lab | — | — | driving-test text and CARLA images | not stated | No | open | — |
| EXPRESS-Bench | 2025 | research-lab | — | — | simulated 3D exploration | not stated | No | open | — |
| HA-VLN 2.0 / HAPS 2.0 | 2025 | research-lab | — | — | simulators with multi-human interactions | not stated | No | open | — |
| HaL-13k (AeroDuo) | 2025 | research-lab | — | — | paired high and low altitude UAVs | not stated | No | open | — |
| IndoorUAV | 2025 | research-lab | — | — | simulated UAV in Habitat indoor scenes | not stated | No | open | — |
| OpenFly | 2025 | research-lab | — | — | Unreal Engine, GTA V, Google Earth, 3D GS renderers | apache-2.0 | Yes | open | — |
| VLN-PE | 2025 | research-lab | — | — | physically embodied robots in simulation | not stated | No | open | — |
| I-FailSense datasets | 2025 | research-lab | — | — | simulation and real-world manipulation | not stated | No | open | — |
| TLA dataset | 2025 | research-lab | — | — | fingertip peg-in-hole assembly with tactile | not stated | No | open | — |
| Box-QAymo | 2025 | research-lab | — | — | Waymo Open | not stated | No | open | — |
| CARScenes | 2025 | research-lab | — | — | images from Argoverse 1, Cityscapes, KITTI, nuScenes | not stated | No | open | — |
| DriveAction | 2025 | research-lab | — | — | production assisted-driving vehicles | not stated | No | open | — |
| DriveBench | 2025 | research-lab | — | — | nuScenes-based frames | not stated | No | open | — |
| DriveLMM-o1 | 2025 | research-lab | — | — | nuScenes multi-view and lidar | not stated | No | open | — |
| Impromptu VLA | 2025 | research-lab | — | — | clips from 8 open driving datasets | not stated | No | open | — |
| LANGTRAJ | 2025 | research-lab | — | — | driving trajectories with language | not stated | No | open | — |
| NuGrounding | 2025 | research-lab | — | — | nuScenes multi-view | not stated | No | open | — |
| NuPlanQA | 2025 | research-lab | — | — | nuPlan multi-view | not stated | No | open | — |
| NuRisk | 2025 | research-lab | — | — | nuScenes, Waymo, CommonRoad | not stated | No | open | — |
| NuScenes-SpatialQA | 2025 | research-lab | — | — | nuScenes | not stated | No | open | — |
| Robusto-1 | 2025 | research-lab | — | — | dashcam video, Peru | not stated | No | open | — |
| V2V-QA (V2V-LLM) | 2025 | research-lab | — | — | V2V4Real multi-vehicle lidar | not stated | No | open | — |
| VLADBench | 2025 | research-lab | — | — | images from public driving sets | not stated | No | open | — |
| WaymoQA | 2025 | research-lab | — | — | Waymo multi-view | not stated | No | open | — |
| Spatial Relationship Aware Dataset for Robotics | 2025 | research-lab | — | — | robot-acquired indoor images | not stated | No | open | — |
| Large-Scale LiDAR-Inertial backpack dataset | 2025 | research-lab | — | — | backpack with multi-beam LiDAR, IMU, RTK-GNSS | not stated | No | open | — |
| Lidar Variability | 2025 | research-lab | — | — | solid-state and spinning lidars | not stated | No | open | — |
| LoopDB | 2025 | research-lab | — | — | camera, loop closure | not stated | No | open | — |
| SLABIM | 2025 | research-lab | — | — | SLAM sensor rig coupled with BIM | not stated | No | open | — |
| SmartPNT-MSF | 2025 | research-lab | — | — | multi-sensor positioning rig | not stated | No | open | — |
| CU-Multi | 2025 | research-lab | — | — | multiple robots | not stated | No | open | — |
| i2Nav-Robot | 2025 | research-lab | — | — | omnidirectional wheeled robot with solid-state LiDARs, 4D radar, stereo, GNSS, IMU | not stated | No | open | — |
| INDOOR-LiDAR | 2025 | research-lab | — | — | robot-centric 360 degree indoor LiDAR, real plus simulated | not stated | No | open | — |
| M2UD | 2025 | research-lab | — | — | ground robot on uneven terrain | not stated | No | open | — |
| M3DGR | 2025 | research-lab | — | — | ground robot multi-sensor SLAM | not stated | No | open | — |
| MobileOcc | 2025 | research-lab | — | — | mobile robot | not stated | No | open | — |
| MRCD (Mobile Robot Campus Dataset) | 2025 | research-lab | — | — | wheeled delivery robot | not stated | No | open | — |
| NavDP dataset | 2025 | synthetic | — | — | simulated navigation | not stated | No | open | — |
| PeRoI | 2025 | research-lab | — | — | mobile robot among pedestrians | not stated | No | open | — |