
Key features
43 DOF
Up to 43 joints, dexterous hands
100 TOPS AI
NVIDIA Jetson Orin NX option
3D LiDAR + Depth
Livox Mid-360, RealSense D435i
2 m/s Walking
120 N·m peak knee torque
Open SDK
Python and C++ for development
About this product
The Unitree G1 EDU is a compact humanoid research and development platform built for universities, robotics labs, and enterprise R&D teams. Standing roughly 1.32 m tall at about 35 kg, it delivers up to 43 degrees of freedom across legs, waist, arms, and optional dexterous hands, with industrial-grade joints reaching 120 N·m of peak knee torque and a stable bipedal walking speed of up to 2 m/s. Its quick-release 9,000 mAh battery supports approximately two hours of runtime and swaps in seconds for continuous experimentation.
Engineered as an open platform, the G1 EDU pairs an 8-core high-performance CPU with an optional NVIDIA Jetson Orin NX (100 TOPS) for on-board AI, plus 3D LiDAR (Livox Mid-360) and an Intel RealSense depth camera for full perception. Open Python and C++ SDKs with imitation- and reinforcement-learning support make it ideal for locomotion research, manipulation, and embodied-AI deployment across EMEIA enterprise and academic programs.
Resources & downloads
Official documentation, SDKs and developer references for this product.
- Product documentationAvailable on request →Manuals and detailed specification documents on request
Developer reference
Core SDKs
Simulation & ecosystem
- unitree_mujocoMuJoCo simulation with sim-to-real support (C++/Python)
- unitree_sim_isaaclabHigh-fidelity Isaac Lab simulation with the robot's DDS interface
- unitree_rl_labReinforcement-learning training environments (Isaac Lab)
- xr_teleoperateTeleoperation via XR headsets (Vision Pro, PICO, Quest) with data recording
- unitree_lerobotEnd-to-end imitation-learning pipeline based on LeRobot
- URDF model (G1_description)Official robot description for simulation and planning
Ready to talk robots?
Tell us about your application and we'll match the right Unitree platform — or design a custom integration around it.
