Image:
ROBOTIS
A semi-humanoid mobile manipulator with a swerve drive and dual arms, optimized for industrial tasks.
NVIDIA Jetson AGX Orin, ROS 2
Height
162.3
cm
Weight
90
kg
Speed
N/A
m/s
Payload
6
kg
Actuators
ROBOTIS DYNAMIXEL-Y series actuators
DoF (Domains of freedom)
25
°
Direct policy training through human task demonstration.
Optimized software stack for seamless robotics development.


High-precision industrial-grade actuators with haptic feedback.
Coordinated 6kg payload manipulation for assembly research.
A semi-humanoid mobile manipulator with a swerve drive and dual arms, optimized for industrial tasks.
At 162.3 cm and 90 kg, this platform features 25 degrees of freedom. It is powered by an NVIDIA Jetson AGX Orin 32GB and includes high-precision DYNAMIXEL-Y actuators, making it a robust environment for Physical AI development.
Image:
ROBOTIS
Specially built for researchers, it supports seamless end-to-end learning. Notable features include high-fidelity haptic feedback for remote operation, multi-modal perception for object tracking, and a 6 kg dual-arm payload capacity.
AI Worker by ROBOTIS
Actuators
ROBOTIS DYNAMIXEL-Y series actuators
DoF (Domains of freedom)
25
°
Height
162.3
cm
Speed
N/A
m/s
Weight
90
kg
Payload
6
kg
Runtime
N/A
h
OS / AI System
NVIDIA Jetson AGX Orin, ROS 2
Built on the proprietary DYNAMIXEL-Y actuator series, which offers high-precision position and current control. The software ecosystem is fully optimized for ROS 2, Python, and C++, providing a direct path from simulation to real-world deployment.
Image:
ROBOTIS
Specifically used by AI developers and university researchers focused on Physical AI and imitation learning. Customers include institutions looking for a "ready-to-code" humanoid platform that bridges the gap between simulation and reality.
Represents a massive leap from the OP-series research bots by moving to the high-torque DYNAMIXEL-Y industrial actuators, providing the payload capacity and precision required for enterprise-level workplace automation tasks.
Specifically designed to learn tasks by observing human demonstrations, allowing it to adopt "human" methods for manipulation and workflow execution.
Uses high-precision actuator feedback to ensure that its movements remain controlled and predictable, essential for building trust during collaboration.
The upper-body configuration and dual-arm setup are modeled after human reach and dexterity to facilitate the study of human-centric tasks.

