DYNAMIXEL-Q

DYNAMIXEL-Q is ROBOTIS’ upcoming QDD actuator platform for humanoids, dynamic robotics, and next-generation Physical AI systems.

Built around Quasi-Direct Drive architecture, DYNAMIXEL-Q is designed for high backdrivability, low impedance, precise torque control, and real-world motion where contact, balance, and compliance matter.

Quasi-Direct Drive design for dynamic, compliant, and torque-aware robotic motion.

Designed for precise torque control in humanoids, balancing robots, and legged systems.

Supports natural interaction, safer contact behavior, and dynamic physical movement.

Built for sim-to-real, imitation learning, reinforcement learning, and real-world deployment.

Built for Real-World Physical AI

Physical AI needs robots that can move through contact, recover from disturbances, and learn from real dynamics — not just simulation. DYNAMIXEL-Q is being developed as a QDD actuator platform for robots that need backdrivable, torque-aware, and compliant movement.

• Designed for dynamic humanoid motion and balance control
• Supports torque-aware behavior for contact-rich robotics
• Helps reduce the gap between simulation and real hardware
• Built for imitation learning and reinforcement learning workflows
• Used in AI Sapiens K0, ROBOTIS’ open-source humanoid Physical AI platform

From Teleoperation to Real Robot Learning

AI Sapiens demonstrates the kind of workflow DYNAMIXEL-Q is designed to support: collect demonstrations, train policies, deploy them to real hardware, and continue improving through real-world robot behavior.

• Leader-follower teleoperation for demonstration capture
• Data collection, training, and inference workflows
• Simulation and real-world deployment for sim-to-real iteration
• Humanoid motion powered by DYNAMIXEL-Q actuators

DYNAMIXEL-Q Preview Models

Early DYNAMIXEL-Q references include QM-060 and QM-080 actuators used in AI Sapiens K0.