Datasets

Browse a curated catalog of datasets for humanoid robots and embodied AI, spanning real-world demonstrations, teleoperation, motion capture, egocentric vision, simulation, manipulation, locomotion, and cross-embodiment robot learning.

83 Datasets · Page 7 of 7

HumanPlus humanoid manipulating a shoe while seated

HumanPlus task data

HumanPlus task data consists of real-robot whole-body demonstrations collected by visually shadowing a nearby human on a customized 33-DoF Unitree H1 with two six-DoF hands. The public download contains warehouse unloading, sweatshirt folding, object rearrangement, and robot greeting demonstrations; the paper additionally reports shoe wearing followed by walking and typing, which are not present in the linked release. Released records support behavior cloning from binocular egocentric RGB and humanoid body and hand targets; the broader system learns a retargeted controller from 40 hours of AMASS motion.

View dataset →

DreamDojo robots interacting with objects across varied real-world environments

DreamDojo GR-1 post-training data

The public DreamDojo package provides 74.3 GB and roughly 7.55 million rows of real Fourier GR-1 manipulation data plus evaluation sets for adapting and testing a robot world model. It contains RGB sequences, robot state, relative actions, end-effector information, task descriptions, and coarse and fine human-action annotations for object pickup, placement, transfer, and other interactions. The release is specifically for target-embodiment post-training and evaluation; it does not contain DreamDojo’s separate 44K-hour human-video pretraining corpus.

View dataset →

Unitree G1 humanoid sorting fruit between colored plates

NVIDIA GR00T Teleop G1

NVIDIA’s GR00T Teleop G1 dataset contains 1,000 real teleoperation trajectories of a Unitree G1 with tri-finger hands performing language-conditioned fruit pick-and-place. Four subsets cover apples, pears, grapes, and starfruit, with upper-body control used to move the requested fruit to a target receptacle. Each trajectory includes 640×480 RGB video at 20 FPS plus 43-D full-body-and-hand state and action vectors in MP4 and HDF5, serving as reference data for GR00T fine-tuning.

View dataset →

Leju Kuavo humanoid and VR operator folding clothing

RoboCOIN humanoid subsets

RoboCOIN contains 180K+ real-world bimanual demonstrations covering 421 tasks, 16 residential, commercial, and workplace scenarios, and 15 robot platforms. Its humanoid subsets include the dexterous-hand Unitree G1edu-u3, collected through motion capture or an exoskeleton, and Leju Kuavo 4 Pro, collected through VR teleoperation. Multiview RGB, joint and end-effector state, gripper articulation, and hierarchical trajectory-, subtask-, and frame-level annotations support standardized multi-embodiment bimanual learning.

View dataset →

RoboMIND robots performing cup handling, wiping, drawer, and toaster tasks

RoboMIND 2.0

RoboMIND 2.0 comprises 310K+ real dual-arm trajectories totaling over 1,000 hours across six embodiments; its project page and abstract report 739 tasks, while the paper body and tables report 759. The embodiments include the bipedal Tien Kung humanoid and wheeled Tian Yi platform. Unified teleoperation produced 20K mobile-manipulation and 12K tactile-enhanced episodes, with RGB-D or multiview RGB, proprioception, actions, and fine-grained natural-language annotations. An additional 20K digital-twin simulation trajectories supports sim-to-real research, long-horizon bimanual learning, contact-rich manipulation, and hierarchical VLA training.

View dataset →

Vision Pro operator teleoperating a Fourier humanoid using a power tool

Fourier ActionNet

Fourier ActionNet contains more than 30K VR-teleoperated trajectories, or roughly 140 hours, from GR1-T1, GR1-T2, and GR2 humanoids using 6-DoF or 12-DoF dexterous hands. It emphasizes bimanual tabletop pick-and-place, pouring, insertion, cabinet interaction, and precise placement with varied household, office, food, and tool objects. RGB-D video, full robot and hand states and actions, end-link poses, and manually verified VLM-generated prompts support imitation learning, VLA, and world-model training.

View dataset →

Unitree G1 performing bimanual cup-and-can manipulation beside an operator

OpenHLM-data

OpenHLM-data is a 298 GB, approximately 1.81-million-row collection of Unitree G1 demonstrations packaged as five corresponding LeRobot datasets. It combines joint-based whole-body teleoperation, stationary manipulation, and robot-free HuMI demonstrations for 12 language-conditioned tasks and fruit-arrangement experiments, including walking, squatting under shelves, pedal use, cart pushing, pouring, and cross-height placement. Stereo head and wrist RGB, 32-D body state, gripper state, actions, and instructions train whole-body-native vision-language-action policies.

View dataset →

Multiple Unitree humanoids performing kitchen and cart-handling tasks

Ψ₀ / psi-data

The 422 GB psi-data repository combines nine released real-world Unitree G1 task archives with SIMPLE simulation and evaluation packages and supporting Humanoid Everyday and retargeting assets. Robot-specific demonstrations use Dex3-1 hands and cover long-horizon activities such as cart pushing, pouring, cleaning, faucet operation, squatting, carrying, and handing objects to people. RGB observations, robot state, whole-body actions, and teleoperation metadata support Ψ₀ fine-tuning and simulation-based evaluation of dexterous loco-manipulation.

View dataset →

Unitree G1 humanoid picking up a colored block

Unitree UnifoLM-WBT

UnifoLM-WBT is Unitree’s rolling collection of task-specific, real-world whole-body teleoperation datasets for the G1 humanoid, comprising 14 releases in the referenced official collection. Tasks include loading plates into a dishwasher, putting clothing into baskets or washing machines, making a bed, and transporting objects while walking. The LeRobot datasets provide root and 29-joint configurations, end-effector and hand states, desired whole-body actions, and task video, targeting whole-body household policy training.

View dataset →

Humanoid robot performing a tabletop tool-use task

Humanoid Everyday

Humanoid Everyday provides 10.3K Unitree G1 and H1 trajectories, exceeding three million frames across 260 tasks. Human-supervised teleoperation records full-body locomotion, dexterous manipulation, tool use, deformables, articulated objects, precision tasks, and human–humanoid interaction in indoor and outdoor environments. Its 30 Hz observations include RGB, depth, LiDAR, tactile feedback, IMU, joint poses and actions, and natural-language descriptions, supporting general-purpose humanoid policy learning and standardized cloud evaluation.

View dataset →

Unitree G1 humanoid manipulating household items on a kitchen table

HIW-500: Humanoids In-the-Wild

HIW-500 contains 500+ hours and 23,743 episodes of human whole-body teleoperation on Unitree G1 robots across 12 real homes in Southeast Asia. Its 11 long-horizon household tasks include room cleanup, table setting, refrigerator restocking, sweeping, and laundry. Synchronized head and wrist cameras, 29-DoF joint state, end-effector state, IMU, odometry, actions, language, and 148K+ fine-grained subtask annotations support imitation learning and general-purpose household mobile manipulation.

View dataset →