H1

Source-backed capabilities

• Operating environments: PIM physical terrain research course, OmniH2O physical imitation research station, OmniH2O human-interaction research demonstration, OmniH2O six-task VR collection station
• Control modes: Autonomous, Teleoperated
• Pim stairs

  Research demonstration - Climb stairs with 15-centimetre risers. - environment: PIM physical terrain research course

  Source
• Pim platform gaps

  Research demonstration - Jump onto platforms and across gaps in a research course. - environment: PIM physical terrain research course

  Source
• Omni dataset

  Research demonstration - Record RGB-D views, head and hand goals, and motor joint targets during VR demonstrations. - environment: OmniH2O six-task VR collection station

  Source
• Omni catch release

  Research demonstration - Catch a red box and release it into a trash bin. - environment: OmniH2O physical imitation research station

  Source
• Omni hammer catch

  Research demonstration - Reach into a box and grasp a hammer with the right hand. - environment: OmniH2O physical imitation research station

  Source
• Omni squat bar

  Research demonstration - Squat when an approaching horizontal bar is lower than head height. - environment: OmniH2O physical imitation research station

  Source
• Omni rock paper scissors

  Research demonstration - Respond to a person’s rock-paper-scissors gesture with a winning hand gesture. - environment: OmniH2O physical imitation research station

  Source
• Omni greeting

  Research demonstration - Select and perform greeting gestures in response to a person’s pose. - environment: OmniH2O human-interaction research demonstration

  Source

Supply / Components

• H1-2

  Core I5 I7 For H1 2 - Compute & Control - Intel - Unitree H1-2 lists Intel Core i5 and Core i7 computers as its standard compute configuration - source: unitree.com

  Source
• H1

  Core I5 I7 For H1 - Compute & Control - Intel - Unitree still lists H1; its current specification names Intel Core i5 and i7 as standard compute - source: unitree.com

  Source
• H1

  Jetson Orin Nx Option For H1 - Compute & Control - NVIDIA - Unitree offers NVIDIA Jetson Orin NX as optional compute for H1 - source: unitree.com

  Source

• Safe Execution of RL Policies via Acceleration-based CBF-QP Constraint Enforcement for Real-World Robotic Deployments

  2026-07-16 - Research Publication - A physical H1 deployment reports 92% fewer constraint violations under the safety layer.

  Source
• Immersive Social Interaction with VR and LLM-Assisted Humanoids

  2026-07-08 - Research Publication - Humanoid robots can extend human presence to remote, constrained, or hazardous environments, but existing teleoperation interfaces often require physically demanding motion tracking or cognitively demanding low-level control. This paper presents an immersive teleoperation framework that integrates voice-controlled locomotion, VR-based manipulation, and bidirectional social interaction for whole-body humanoid control.

  Source
• Rokaia19/h1-mocap-retargeting

  2026-07-05 - Research Publication - Retargeting human motion capture data to the Unitree H1 robot using IK and MuJoCo

  Source
• iitd-bird-robotics/2026-major-Bipedal_Locomotion_Using_RL

  2026-06-29 - Research Publication - RL-based framework for robust bipedal locomotion on uneven terrains and staircases using the Unitree H1 humanoid robot in the Genesis simulator with PPO, curriculum learning, gait phase generation, and domain randomization.

  Source
• Open-X-Humanoid/HEX

  2026-05-19 - Research Publication - HEX is a whole-body vision-language-action framework for full-sized humanoid robots.

  Source
• Constraint-Enhanced Reinforcement Learning Based on Dynamic Decoupled Spherical Radial Squashing

  2026-05-05 - Research Publication - When deploying reinforcement learning policies to physical robots, actuator rate constraints -- hard limits on how fast each joint can move per control step -- are unavoidable.

  Source
• FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control

  2026-04-06 - Research Publication - FlashSAC scales off-policy reinforcement learning with larger networks, high-throughput data and bounded optimization norms. It benchmarks diverse simulation tasks and transfers blind locomotion policies to a physical 29-DoF Unitree G1, including flat motion and stair climbing.

  Source
• Embedding Classical Balance Control Principles in Reinforcement Learning for Humanoid Recovery

  2026-03-09 - Research Publication - This balance-informed reinforcement-learning study combines capture-point, center-of-mass and momentum signals during training, then transfers a proprioceptive policy to physical Unitree H1-2. Ten hardware trials reached upright stance, with safety intervention after recovery in three; the reported 93.4% recovery rate comes from simulation.

  Source
• Residual Control for Fast Recovery from Dynamics Shifts

  2026-03-08 - Research Publication - Robotic systems operating in real-world environments inevitably encounter unobserved dynamics shifts during continuous execution, including changes in actuation, mass distribution, or contact conditions. When such shifts occur mid-episode, even locally stabilizing learned policies can experience substantial transient performance degradation. While input-to-state stability guarantees bounded state deviation, it does n

  Source
• Embodiment-Aware Generalist Specialist Distillation for Unified Humanoid Whole-Body Control

  2026-02-03 - Research Publication - Humanoid Whole-Body Controllers trained with reinforcement learning (RL) have recently achieved remarkable performance, yet many target a single robot embodiment.

  Source
• FRoM-W1: Towards General Humanoid Whole-Body Control with Language Instructions

  2026-01-19 - Research Publication - Humanoid robots are capable of performing various actions such as greeting, dancing and even backflipping.

  Source
• AIM-Intelligence/video2robot

  2025-12-18 - Research Publication - End-to-end pipeline converting generative videos (Veo, Sora) to humanoid robot motions

  Source
• A Hierarchical Framework for Humanoid Locomotion with Supernumerary Limbs

  2025-11-25 - Research Publication - The integration of Supernumerary Limbs (SLs) on humanoid robots poses a significant stability challenge due to the dynamic perturbations they introduce.

  Source
• HansZ8/RoboJuDo

  2025-09-28 - Research Publication - A plug-and-play deploy framework for robots. Just deploy, just do.

  Source
• RobotDancing paper validates long-horizon humanoid motion tracking on Unitree H1

  2025-09-25 - Research Publication - Long-horizon, high-dynamic motion tracking on humanoids remains brittle because absolute joint commands cannot compensate model-plant mismatch, leading to error accumulation.

  Source
• RoMoCo: Robotic Motion Control Toolbox for Reduced-Order Model-Based Locomotion on Bipedal and Humanoid Robots

  2025-09-23 - Research Publication - We present RoMoCo, an open-source C++ toolbox for the synthesis and evaluation of reduced-order model-based planners and whole-body controllers for bipedal and humanoid robots.

  Source
• Latent Conditioned Loco-Manipulation Using Motion Priors

  2025-09-19 - Research Publication - Although humanoid and quadruped robots provide a wide range of capabilities, current control methods, such as Deep Reinforcement Learning, focus mainly on single skills. This approach is inefficient for solving more complicated tasks where high-level goals, physical robot limitations and desired motion style might all need to be taken into account. A more effective approach is to first train a multipurpose motion pol

  Source
• Karthus-Chen/unitree_h1_learn

  2025-09-12 - Research Publication - A robot simulation environment based on mujoco, supporting research on teleoperation and imitation learning

  Source
• Yara-NM/h1_unitree_arms_controller

  2025-07-23 - Research Publication - Low- and high-level controllers for Unitree H1 robot arms with dual-arm manipulation support.

  Source
• JeffrinSam/JeffrinSam_G1

  2025-07-22 - Research Publication - Synthetic Data Generation -Generate synthetic joint data for Unitree H1/G1 robots via imitation learning in Isaac Lab. Simulate and visualize in Isaac Sim, enhance with Cosmos for realism, and combine with teleop data to fine-tune Gr00t N1.5 for pick-and-place tasks, tested in Isaac Sim(Inference)

  Source
• Galbot open-sources OpenWBT whole-body teleoperation for humanoid robots

  2025-06-09 - Research Publication - This repo covers Galbot open-sources OpenWBT whole-body teleoperation for humanoid robots for Galbot G1 in Human-robot interaction.

  Source
• Yara-NM/MPC-h1-Manipulation-task

  2025-05-28 - Research Publication - This repository implements a two-layer Model Predictive Control (MPC) system for motion control of the Unitree H1 robot’s arms. The control architecture includes a high-level Trajectory MPC for planning optimal end-effector paths and a low-level Kinematics MPC for computing feasible joint motions. using a reduced 8-DOF arms model

  Source
• GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning

  2025-05-24 - Research Publication - Recent advances in reinforcement learning (RL) have demonstrated the powerful exploration capabilities and multimodality of generative diffusion-based policies.

  Source
• TD-GRPC: Temporal Difference Learning with Group Relative Policy Constraint for Humanoid Locomotion

  2025-05-19 - Research Publication - Robot learning in high-dimensional control settings, such as humanoid locomotion, presents persistent challenges for reinforcement learning (RL) algorithms due to unstable dynamics, complex contact interactions, and sensitivity to distributional shifts during training.

  Source
• Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning

  2025-04-19 - Research Publication - ALMI coordinates separately trained lower-body locomotion and upper-body motion imitation on physical Unitree H1-2. Its companion ALMI-X collection contains language-annotated MuJoCo trajectories and supports released whole-body control models.

  Source
• Humanoid Agent via Embodied Chain-of-Action Reasoning with Multimodal Foundation Models for Zero-Shot Loco-Manipulation

  2025-04-13 - Research Publication - Humanoid-COA decomposes natural-language instructions into grounded action sequences and evaluates physical H1-2 and G1 loco-manipulation in an open area and apartment. The framework combines visual scene interpretation, affordance reasoning, manipulation primitives and navigation.

  Source
• InternRobotics/HoST

  2025-04-03 - Research Publication - [RSS 2025 Best Systems Paper Finalist] 💐Official implementation of "Learning Humanoid Standing-up Control across Diverse Postures"

  Source
• Humanoid Whole-Body Locomotion on Narrow Terrain via Dynamic Balance and Reinforcement Learning

  2025-02-24 - Research Publication - DBHL learns whole-body locomotion with a dynamic-balance reward and proprioceptive control. Physical H1-2 experiments traverse a narrow track, stairs and uneven ground while handling payloads, pushes and trips, separately from simulated success-rate comparisons.

  Source
• A Unified and General Humanoid Whole-Body Controller for Versatile Locomotion

  2025-02-05 - Research Publication - Locomotion is a fundamental skill for humanoid robots. However, most existing works make locomotion a single, tedious, unextendable, and unconstrained movement.

  Source
• LeCAR-Lab/HumanoidVerse

  2025-01-28 - Research Publication - ain_rand.push_robots=True \ +simulator=genesis # you can load the policy trained in isaacgym ``` # Start Training Your Humanoids! Here are some starting commands to train & evaluate the locomotion policy on Unitree H1 Humanoid Robot among multiple simulators. ## IsaacGym Training Command ```bash python humanoidverse/train_agent.py \ +simulator=isaacgym \ +exp=locomotion \ +domain_rand=NO_domain_rand \ +rewards=loco/reward_h1_locomotion \ +robot=h1/h1_10dof \ +te

  Source
• NVlabs/HOVER

  2024-12-19 - Research Publication - HOVER

  Source
• DepthAnything/PromptDA

  2024-12-17 - Research Publication - [CVPR 2025] Prompt Depth Anything

  Source
• kyavuzkurt/h1_description

  2024-12-03 - Research Publication - ROS2 Package for Unitree H1 Robot Description

  Source
• Learning Humanoid Locomotion with Perceptive Internal Model

  2024-11-21 - Research Publication - In contrast to quadruped robots that can navigate diverse terrains using a "blind" policy, humanoid robots require accurate perception for stable locomotion due to their high degrees of freedom and inherently unstable morphology. However, incorporating perceptual signals often introduces additional disturbances to the system, potentially reducing its robustness, generalizability, and efficiency.

  Source
• HOVER unifies multiple whole-body control modes for humanoid robots

  2024-10-28 - Research Publication - HOVER introduces a multi-mode policy distillation framework for navigation, loco-manipulation, and tabletop manipulation control modes. The paper uses full-body kinematic motion imitation as a shared abstraction so one policy can move between different command spaces.

  Source
• LeCAR-Lab/human2humanoid

  2024-09-29 - Research Publication - [IROS 2024] Learning Human-to-Humanoid Real-Time Whole-Body Teleoperation. [CoRL 2024] OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning

  Source
• Structural Optimization of Lightweight Bipedal Robot via SERL

  2024-08-28 - Research Publication - Designing a bipedal robot is a complex and challenging task, especially when dealing with a multitude of structural parameters.

  Source
• ZhangXG001/unitree_h1_humanoid-gym

  2024-08-07 - Research Publication - an implement of unitree_h1 robot with humanoid-gym

  Source
• fan-ziqi/robot_lab

  2024-07-06 - Research Publication - The source registers exact model-specific Isaac Lab environments for tracked quadruped, wheeled-quadruped and humanoid platforms.

  Source
• OmniH2O enables whole-body teleoperation and autonomy on Unitree H1

  2024-06-13 - Research Publication - OmniH2O presents a learning-based system for whole-body humanoid teleoperation and autonomy using kinematic pose as a control interface. The work demonstrates real-world whole-body tasks through teleoperation and autonomy, including sports, object movement, and human interaction.

  Source
• carlosferrazza/humanoid-bench

  2024-03-18 - Research Publication - /dreamerv3 * TD-MPC2: https://github.com/nicklashansen/tdmpc2 * purejaxrl (JAX-PPO traning): https://github.com/luchris429/purejaxrl/tree/main * Digit models: https://github.com/adubredu/KinodynamicFabrics.jl/tree/sim * Unitree H1 models: https://github.com/unitreerobotics/unitree_ros/tree/master * MuJoCo Menagerie (Unitree H1, Shadow Hands, Robotiq 2F-85 models): https://github.com/google-deepmind/mujoco_menagerie * Robosuite (some texture files): https://github.com/ARISE-Initiative/robosuite

  Source
• External H2O paper demonstrates real-time whole-body teleoperation on Unitree H1

  2024-03-07 - Research Publication - We present Human to Humanoid (H2O), a reinforcement learning (RL) based framework that enables real-time whole-body teleoperation of a full-sized humanoid robot with only an RGB camera.

  Source
• SelfBriefs/PBRS-H1

  2024-03-04 - Research Publication - PBRS on Unitree H1 Humanoid Robot

  Source
• rohanpsingh/LearningHumanoidWalking

  2022-07-13 - Research Publication - Training a humanoid robot for locomotion using Reinforcement Learning

  Source
• unitreerobotics/unitree_mujoco

  Research Publication - Unitree’s MuJoCo simulator integrates low-level SDK2 control with C++ and Python simulation. It supplies robot models, joint-command/state interfaces, terrain tools and controller examples for digital validation.

  Source
• Unitree robot descriptions and simulation packages

  Research Publication - Unitree's shared robot-description and simulation repository publishes robot models, meshes and ROS control packages for multiple platforms.

  Source
• Unitree SDK2 control interfaces and examples

  Research Publication - Unitree SDK2 supplies C++ communication libraries and platform-specific examples for robot control. The source separates G1, H1, H2 and R1 examples alongside quadrupeds and shared DDS interfaces.

  Source

Media library

• H1 impact-recovery demonstration

  image · Unitree Robotics

  Source
• H1 perception coverage

  image · Unitree Robotics

  Source

Official product specifications, downloads, developer documentation and exact-product media for the Unitree H1 humanoid robot.

• H1 download and manual centre

  WEB · H1 · Unitree Robotics

  Source

• Unitree reports 10 m/s sprint with modified H1

  2026-04-11 - Demonstration - The test configuration weighed approximately 62 kg, with the head and hands removed; it differed from the standard 47 kg H1. The run took place during testing ahead of Beijing’s April 19 humanoid half-marathon.

  Source
• Balance-informed learning study tests H1-2 fall recovery

  2026-03-09 - Research Publication - A physical H1-2 recovered from supine, seated, kneeling and crouched starts in all 10 trials, using an unchanged 50 Hz policy. Three trials developed joint oscillations after recovery and prompted safety intervention; the reported 93.4% aggregate success figure came from simulation.

  Source
• EAGLE study transfers control across four humanoids

  2026-02-03 - Research Publication - A policy trained in simulation controlled physical Unitree H1, Unitree G1, Booster T1 and Fourier N1 robots without hardware-specific fine-tuning. Hardware demonstrations covered walking, leaning and squatting; a fifth platform, PNDbotics Adam, was included in simulation.

  Source
• FRoM-W1 study tests language-driven H1 and G1 control

  2026-01-19 - Research Publication - Physical deployments used H1 with Inspire hands and a torso ZED Mini camera, and a 29-DoF G1 EDU+ with Dex3-1 hands. A separate workstation ran the language model, while onboard computers executed control; the G1 waist pitch and roll joints were locked for stability.

  Source
• Humanoid Everyday releases G1 and H1 manipulation data

  2025-10-09 - Research Publication - The initial paper described 10,300 trajectories and more than three million frames spanning 260 tasks in seven categories. Collection used a 29-DoF G1 with Dex3-1 hands and a 27-DoF H1 configuration with Inspire hands; RGB-D, lidar and robot-state streams accompanied demonstrations.

  Source
• RobotDancing study tests long-horizon humanoid tracking

  2025-09-25 - Research Publication - Physical G1 tests completed several LAFAN1 dance sequences; three sequences were untested because of robot maintenance. H1 and H1-2 demonstrations extended the training recipe to larger hardware, where torque-to-weight limits constrained extreme movement.

  Source
• ALMI study demonstrates H1-2 whole-body coordination

  2025-04-19 - Research Publication - The physical H1-2 ran inference at 50 Hz on an onboard Jetson Orin NX, combining velocity-commanded movement with upper-body tracking. VR wrist and finger capture enabled separate teleoperated loco-manipulation trials; the released MuJoCo trajectory dataset was simulated.

  Source
• Chain-of-action study tests H1-2 planning

  2025-04-13 - Research Publication - The physical 27-DoF H1-2 used Inspire hands and two RGB-D cameras, with separate views for manipulation and navigation. Tests covered carrying a box to a table, placing a bottle in a box and collecting toys, using inverse kinematics and high-level navigation commands.

  Source
• DBHL study tests H1-2 balance on narrow terrain

  2025-02-24 - Research Publication - Physical tests used a 25 cm wooden track with a ramp, bridge and stairs, relying on proprioception rather than vision or lidar. Demonstrations included carrying 5 kg, human pushes, conical obstacles and a moving stick; separate trials traversed irregular planar obstacles.

  Source
• HugWBC study validates H1 whole-body control

  2025-02-05 - Research Publication - Hardware tests measured body-pitch and waist-rotation tracking during standing, walking and jumping, with five trials per command and gait. The controller allows body height, foot swing and gait settings to vary while accepting external upper-body commands.

  Source
• Unitree H1 robots perform Yangko at 2025 Spring Festival Gala

  2025-01-28 - Demonstration - The Zhang Yimou-directed routine paired the robots with 16 dancers in a traditional Northeast Chinese folk dance. The robots underwent three months of rehearsal and AI training, with laser SLAM supporting positioning and formation changes across gaps in the stage floor.

  Source
• Perceptive Internal Model study tests H1 terrain traversal

  2024-11-21 - Research Publication - Physical H1 experiments included 15 cm stairs and stepping onto a wooden platform before crossing a gap. The study compared lidar-based mapping with a separate camera-based configuration and also evaluated Fourier GR-1.

  Source
• HOVER study validates multimode H1 control

  2024-10-28 - Research Publication - The physical platform was a 19-DoF H1; quantitative tests tracked 20 standing motions and compared control modes with specialist policies. Additional locomotion demonstrations tested switching among command interfaces; automatic mode selection remained future work.

  Source
• OmniH2O study demonstrates H1 teleoperation and learning

  2024-06-13 - Research Publication - The OmniH2O-6 dataset contains about 40 minutes of physical teleoperation across six tasks, pairing RGB-D images, head-and-hand goals and motor targets at 30 Hz. Separate experiments learned four tasks from demonstrations and used GPT-4o to select motion primitives; sports showcases included human teleoperation.

  Source
• Unitree H1 demonstrates standing backflip without hydraulics

  2024-03-20 - Demonstration - The robot used in-house M107 electric joint motors and reinforcement-learning simulation to train the maneuver. The landing included a small corrective hop; the approximately 1.8 m, 47 kg platform had previously demonstrated stair climbing, jumping and speed-walking.

  Source
• H2O study demonstrates H1 whole-body teleoperation

  2024-03-07 - Research Publication - Physical demonstrations included kicking a ball, pushing a stroller, catching and moving a box, and coordinated locomotion. A simulation-trained motion imitator used filtered, retargeted human motion data before zero-shot transfer to the robot.

  Source
• Unitree H1 demonstrates claimed 3.3 m/s speed

  2024-03-01 - Demonstration - The same video showed dancing, stair ascent and descent, and a standing jump. The electrically driven robot used a head-mounted depth camera and lidar; Unitree framed the measured speed as a full-size humanoid record.

  Source
• Unitree introduces its first H1 humanoid robot

  2023-08-15 - Product Launch - The electric platform uses five degrees of freedom per leg and knee joints rated at up to 360 Nm. At the subsequent World Robot Conference, Unitree showed a video of H1 recovering its balance after kicks and described future industrial and consumer applications.

  Source

Market context

Full-size general-purpose humanoid with lidar and depth vision.

• Target environment: Research labs, university robotics programs and developer workshops.
• Workflow context: Locomotion research, bipedal control experimentation and embodied AI prototyping at full human scale.
• Customer context: Robotics researchers, AI labs and developers seeking a full-size research humanoid.
• Deployment model: Commercial product sold as a development platform rather than a workplace automation tool.
• Commercial maturity: Commercial product with broad developer adoption for research use.
• Adoption constraints: Primarily a research platform rather than a deployment-ready industrial system.
• Market position: Sits alongside G1 as Unitree's full-size research humanoid. Serves as a performance benchmark for bipedal locomotion and humanoid control research.
• Adjacent products: G1, AI Sapiens K0

Comparable Systems

• AI Sapiens K0

  ROBOTIS - humanoid robotics

• Booster T1

  Booster Robotics - humanoid robotics

• Booster T2

  Booster Robotics - humanoid robotics

• G1

  Unitree - humanoid robotics

• G1+

  Unitree - humanoid robotics

• LimX Oli

  LimX Dynamics - Humanoid Platform

Related Systems

• Digit

  Agility Robotics - humanoid robotics