Home
/
Robots
/
Fourier GR-2

Fourier GR-2

Fourier GR-2 is a humanoid robot with 53 DOF, 12-DOF tactile hands, 380 Nm actuators, 2-hour battery, and advanced AI for research, industry, and service applications.
In Development
China
Humanoid
2024
Specifications
Dev Doc
Website
Made by
Shanghai Fourier Intelligent Technology Co., Ltd.
Software Type
Closed Source
Software Package
SDK compatible with NVIDIA Isaac Lab, ROS, Mujoco. AI-driven control and perception modules. Support for force-feedback and tactile sensor integration. Cloud-based updates and developer tools.
Actuators
Equipped with FSA 2.0 intelligent actuators delivering peak torques exceeding 380 Nm, providing precise, powerful, and energy-efficient joint control.
Compiute
Integrated high-performance computing modules support AI inference, sensor fusion, and real-time control, enabling complex autonomous behaviors.
Sensors
Includes tactile sensor arrays on hands, force sensors, cameras, and depth sensors for comprehensive perception and adaptive manipulation.
Max Op. time
120
mins

Robot Brief

Fourier GR-2 is the next-generation humanoid robot from Shanghai-based Fourier Intelligence, building on the success of its predecessor, GR-1. Standing 175 cm tall and weighing approximately 63-65 kg, the GR-2 features 53 degrees of freedom, including 12 degrees of freedom in its dexterous hands equipped with tactile sensors for real-time adaptive manipulation. The robot is powered by advanced FSA 2.0 actuators delivering peak torques over 380 Nm, enabling agile and precise movements such as walking, grasping, and complex object handling. Its detachable battery doubles the runtime to about 2 hours and supports quick swapping for prolonged use. Fourier redesigned the internal joint configuration from parallel to serial, improving debugging, reducing manufacturing costs, and enhancing the transition from AI simulation to real-world applications. The GR-2 supports multiple control modes including VR remote control, lead-through programming, and direct command, and is compatible with popular software frameworks like NVIDIA Isaac Lab, ROS, and Mujoco. With an integrated cabling design for compactness and modularity, the GR-2 aims to advance human-robot collaboration in research, industry, and commercial deployments.

Use Cases

GR-2 autonomously performs complex bipedal locomotion and manipulation tasks, including walking at speeds up to 5 km/h and lifting up to 3 kg per arm. Its tactile-sensor-equipped hands can recognize object shapes and materials, enabling adaptive grasping and interaction. The robot supports various programming and control methods, facilitating rapid development and deployment in dynamic environments.

Industries

  • Research & Development: Provides a versatile platform for robotics and AI innovation.
  • Industrial Automation: Supports manufacturing, assembly, and logistics tasks requiring precision.
  • Healthcare & Rehabilitation: Assists with patient support and therapeutic applications.
  • Education: Facilitates robotics learning and experimentation.
  • Service & Hospitality: Enhances customer interaction and service automation.

‍

Specifications

Length
552
mm
Width
275
mm
Height (ResT)
mm
Height (Stand)
1750
mm
Height (Min)
mm
Height (Max)
mm
Weight (With Batt.)
54
kg
Weight (NO Batt.)
63
kg
Max Step Height
mm
Max Slope
+/-
°
Op. Temp (min)
°C
Op. Temp (Max)
°C
Ingress Rating
No items found.

Intro

Fourier GR-2 features a 175 cm tall, 63-65 kg biomimetic body with 53 degrees of freedom, including 12 DOF in each hand equipped with six tactile sensor arrays. The robot’s advanced FSA 2.0 actuators provide peak torques over 380 Nm, enabling dynamic and precise movement. Its detachable battery doubles operational time to approximately 2 hours and allows quick replacement. The internal design utilizes a serial joint configuration and integrated cabling, reducing complexity and improving modularity. GR-2 supports multiple control modes such as VR remote operation, lead-through programming, and direct command, making it developer-friendly. The software platform is compatible with NVIDIA Isaac Lab, ROS, and Mujoco, supporting AI, vision, and force-feedback control.

‍

Connectivity

  • WiFi (2.4 GHz and 5 GHz)
  • Ethernet ports
  • USB ports
  • Cloud connectivity for updates and remote monitoring.

‍

Capabilities

  • Agile bipedal locomotion at speeds up to 5 km/h
  • Dexterous manipulation with 12-DOF tactile-sensor-equipped hands
  • Adaptive grasping and object recognition via force sensors
  • Multiple control modes: VR, lead-through programming, direct commands
  • Modular design for easy maintenance and customization
  • Real-time AI-driven motion planning and environment interaction.

‍