G1 Capabilities

Capability

generalist

Source-backed capabilities

  • Capability areas: Bipedal locomotion
  • Workflows: Experimentation, locomotion research, teleoperation and embodied AI prototyping at a lower cost barrier than full-size humanoids.
  • Operating environments: Research lab
  • Target markets: Robotics researchers, AI labs and developers seeking an accessible humanoid platform.
  • Deployment model: Commercial product sold as a development platform, not as a workplace automation tool.
  • Dynamic locomotion - Demonstrated - 2025-09-25 - Unitree's G1 humanoid robot achieved a running speed of 2 meters per second. - source: arxiv.org
  • Navigation and obstacle avoidance - Demonstrated - 2025-09-09 - Unitree Robotics showcased the G1 humanoid robot's autonomous navigation capabilities at the WOW Summit. - source: auki.com
  • Autonomous route navigation - Demonstrated - 2025-09-09 - Unitree Robotics showcased the G1 humanoid robot's autonomous navigation capabilities at the WOW Summit. - source: auki.com
  • Bipedal walking - Research demonstration - 2026-07-27 - Learning Reusable Hybrid Motion Priors for Humanoid Locomotion from Motion Imitation - source: arxiv.org
  • Multi-terrain locomotion - Research demonstration - 2026-06-09 - The authors validate the approach in simulation and demonstrate successful deployment on a Unitree G1 humanoid robot navigating sparse footholds with lateral constraints. - source: arxiv.org
  • Fall recovery - Research demonstration - 2026-05-18 - The paper introduces a state-dependent adversarial motion prior framework that lets one policy handle walking, running, and fall recovery on Unitree G1. - source: arxiv.org
  • Whole-body loco-manipulation - Research demonstration - 2026-07-20 - Closing the Loop in Humanoid VLA: Persistent 3D Object Tokens for Verifiable Loco-Manipulation - source: arxiv.org
  • Dexterous manipulation - Research demonstration - 2026-06-30 - RoboTacDex: A Dexterous Visual-Tactile-Action Dataset for Humanoid Manipulation - source: arxiv.org
  • Object manipulation - Research demonstration - 2026-06-30 - In the field of robot learning, large-scale and diverse demonstration trajectories provide the fundamental basis for enhancing robotic manipulation ability. - source: arxiv.org
  • Trajectory tracking - Research demonstration - 2026-06-24 - The paper “Learning Asynchronous Upper-body Task-space Trajectory Tracking Policy for Humanoid Robots” was published on June 24, 2026. - source: arxiv.org
  • Pick and place - Research demonstration - 2025-06-01 - NVIDIA GEAR released a Unitree G1 fruits pick-and-place dataset with 1,000 teleoperation trajectories of real robot data. - source: huggingface.co
  • Remote and shared control - Research demonstration - 2026-05-12 - The paper presents a real-time whole-body teleoperation system that maps human motion from a Virdyn IMU-based suit directly onto Unitree G1. - source: arxiv.org
  • Whole-body motion tracking - Research demonstration - 2026-05-12 - The work gives G1 a practical teleoperation benchmark for low-latency whole-body retargeting using wearable motion capture. - source: arxiv.org
  • Dynamic acrobatics - Demonstrated - 2026-02-23 - The robots, dressed in red vests, executed complex maneuvers including backflips, cartwheels, and high-speed kickboxing in unison. - source: news.cgtn.com
  • Backhand hitting - Research demonstration - 2026-06-24 - The paper “TaskNPoint: How to Teach Your Humanoid to Hit a Backhand in Minutes” was published on June 24, 2026. - source: arxiv.org
  • Soccer shooting - Research demonstration - 2026-06-09 - In real world on a Unitree G1 with onboard perception, RoboNaldo attains 0.73 m and 0.86 m average target shooting error from 3 m away in free-kick and moving-ball cases, accordingly. - source: arxiv.org
  • Table-tennis play - Research demonstration - 2026-03-12 - Struck incoming tennis balls and sustained multi-shot rallies on the physical humanoid. - environment: real-world table-tennis research setup - source: arxiv.org
  • Warehouse picking and stowing - Research demonstration - 2026-07-22 - Restocked bags of chips in a real supermarket task after data-efficient post-training. - environment: real supermarket restocking setup - source: arxiv.org

View G1 overview