
Unitree G1 Comp Humanoid Robot
Choose options

Product Description
The Unitree G1-Comp is Unitree's premier athletic humanoid platform, purpose-built for soccer-style robotics challenges and motion-intensive competition environments. Weighing 35 kg with 25–45 degrees of freedom, the G1-Comp delivers explosive starts, stable turning, and rapid impact recovery through 120 N·m of knee torque and reinforced joint architecture.
Its foldable frame collapses from 1,320 × 450 × 200 mm down to 690 × 450 × 300 mm for easy transport between venues, while still supporting on-field agility once deployed. The G1-Comp sprints at up to 3.3 m/s on uneven turf while maintaining ±2° pitch-roll stability, executes fast kicks at ~12 m/s with just 0.5 s windup, and dribbles at speeds up to 2.5 m/s. It can carry 3 kg payloads up 20° inclines for 90 minutes on a single 9,000 mAh charge, with fast-swap batteries minimising downtime between matches.
Perception is handled by a Livox MID-360 LiDAR and Intel RealSense D435 depth camera running YOLO11-based stadium detection for real-time SLAM mapping, target tracking, and opponent localisation. An NVIDIA Jetson Orin NX module delivers 100–157 TOPS of onboard AI compute, and the full software stack includes the unitree_rl_gym reinforcement learning framework, Isaac Gym and MuJoCo simulation pipelines, and a dedicated RoboCup SDK with APIs for motion control, spatial positioning, vision processing, and competition-rule integration.
Key Specifications
- Weight: 35 kg | Degrees of Freedom: 25–45
- Frame: Foldable — 1,320×450×200 mm (deployed) to 690×450×300 mm (folded)
- Knee Torque: 120 N·m | Sprint Speed: 3.3 m/s | Stability: ±2° pitch/roll on uneven turf
- Kick Speed: ~12 m/s (0.5 s windup) | Dribble Speed: up to 2.5 m/s
- Payload: 3 kg up 20° inclines for 90 minutes
- Battery: 9,000 mAh, fast-swap | Runtime: ~2 hours continuous
- Perception: Livox MID-360 LiDAR + Intel RealSense D435, YOLO11 stadium detection
- Compute: NVIDIA Jetson Orin NX, 100–157 TOPS
- Software: unitree_rl_gym, Isaac Gym, MuJoCo, dedicated RoboCup SDK
- Joints: Dual-encoder for enhanced precision
- Use cases: RoboCup-style competition, sports robotics research, sim-to-real AI development