Humanoids

Advanced humanoid platforms engineered for real-world interaction and applied intelligence. Designed for AI research, development, and operational deployment, these systems combine mobility, perception, and autonomy to support next-generation innovation.

UNITREE H2

A full-scale humanoid engineered for high-torque performance and next-generation embodied intelligence.

  • Built with Elegance. Powered by Intelligence.
    Standing 180 cm tall, merging aesthetic form with operational precision.
  • High-Performance Motion Control
    31 degrees of freedom and 360 N·m joint torque, with continuous OTA updates for evolving performance.
  • Embodied Intelligence Core
    2070 TOPS of onboard compute — built for advanced AI research, model training, and complex perception tasks.

UNITREE R1

An ultra-lightweight humanoid platform designed for agile motion, open development, and rapid experimentation.

  • Agile by Design
    ≤123 cm, 29 kg, 20–26 DOF and a 2-DOF articulated head — for dynamic locomotion and fluid environmental adaptation.
  • See, Hear, Act
    Integrated multimodal AI supports voice and visual inputs, connecting perception to action in real-time.
  • Open to What's Next
    Open control interfaces and simulation compatibility for seamless virtual-to-physical deployment.
  • Built to Stand Out
    Modular exterior and customizable finishes adapt R1 to any environment, where form matters as much as function.

UNITREE G1

A versatile humanoid agent built for practical interaction, task execution, and applied AI applications.

  • Extended Joint Range
    Equipped with 23–43 joint motors, expanded posture control and full-body kinematic experimentation.
  • AI-Based Learning Framework
    Supports imitation and reinforcement learning for rapid algorithm testing and behavioral optimization.
  • Force-Control Dexterous Hand
    Force-position hybrid control for stable, precise manipulation — designed to replicate the dexterity of the human hand.
  • Open Model Integration
    Built around UnifoLM (Unified Robot Large Model), G1 supports world-model integration and large-scale embodied AI experimentation.