The GR-2 marks a new milestone in Fourier's GRx series. Building on the data and technology accumulated with the GR-1, the GR-2 delivers comprehensive breakthroughs in hardware design, development framework, and commercialization — featuring more dexterous upper-limb manipulation, a more powerful actuation system, and a more open development platform.

Standing 175 cm tall and weighing 63 kg, the GR-2 features an efficient full-body series-link structure with internal cable routing for a clean aesthetic, modular design for easy maintenance, and a battery life of up to 2 hours. Its 53 degrees of freedom and FSA 2.0 actuators deliver a maximum joint torque of 380 N·m and a walking speed of up to 5 km/h.

The GR-2 is equipped with a next-generation 12-DOF dexterous hand per arm, featuring 6 array-type tactile sensors for precise force-controlled grasping of objects with varying shapes, materials, and weights. Multiple teleoperation modes are supported — including VR remote control, drag-and-drop teaching, and host-side end-effector control — enabling rich real-world training data collection.

Powered by the Fourier Toolkit, the GR-2 offers a comprehensive SDK built on mainstream programming languages, with APIs covering machine vision, path planning, and force-feedback control. The platform supports NVIDIA Isaac Lab, ROS, MuJoCo, Webots, and other open-source frameworks, lowering the barrier for researchers and developers worldwide.

• Height: 175 cm
• Weight: 63 kg
• Battery life: 2 hours
• Total degrees of freedom: 53
• Maximum joint torque: 380 N·m
• Walking speed: up to 5 km/h
• Dexterous hands: 12 DOF per hand
• Tactile sensors: 6 array-type sensors per hand
• Actuator: FSA 2.0 (7 models, dual-encoder system)
• Teleoperation: VR remote control, drag-and-drop teaching, host-side end-effector control
• Supported frameworks: NVIDIA Isaac Lab, ROS, MuJoCo, Webots