GR-2 Humanoid Robot Research Summary

 

 

 

Company Background: Fourier Intelligence

 

Fourier Intelligence is a Shanghai-based technology company founded in 2015 by Alex Gu. The company initially focused on exoskeleton robots and intelligent rehabilitation systems, aiming to leverage full-stack robotics technology to enrich people’s lives. Fourier has since expanded its portfolio to include humanoid robots. In July 2024, Fourier Intelligence rebranded to Fourier and Fourier Rehab, signifying its dedication to expanding its offerings. The company’s vision is to integrate various cutting-edge solutions into a single clinical pathway.

Robot History and Development Timeline: GR-2

 

The GR-2 is part of Fourier’s GRx Humanoid Robot Series, guided by the philosophy of “Made for AI”. While the exact launch date of GR-2 is not explicitly stated in all snippets, news from September 2024 announced that Fourier Unveils the Next-Generation Humanoid Robot GR-2. This indicates a recent development or a significant upgrade from previous models, such as the GR-1, which was announced in July 2023 and slated for mass production. The GR-2 introduces significant advancements, including 12-DoF dexterous hands that double the dexterity of previous models, and is designed to mirror human physiological flexibility. Fourier has been actively training its humanoid robots for real-world roles using NVIDIA Isaac Gym, as highlighted in November 2024. The company anticipates a tenfold increase in the humanoid robot industry, while continuing to produce the GR-2 on a small scale.

Key Technical Specifications

 

The Fourier GR-2 is a robust humanoid robot designed for enhanced agility, precision, and dexterity. Key specifications include:
  • Height: 175 cm
  • Weight: 63 kg
  • Degrees of Freedom (DoF): 53 DoF
  • Hands: 12-DoF dexterous hands with six array-type tactile sensors, enabling force sensing, object shape and material identification, and real-time grip adjustment
  • Battery/Runtime: Approximately 2 hours
  • Peak Torques: 380 N.m
  • Speed: 5 km/h
  • Load Capacity: Single-arm load capacity of 3 kg

 

 

 

 

AI/Software Stack

 

The GR-2 is designed with the philosophy of “Made for AI”. Fourier trains its humanoid robots for real-world roles using NVIDIA Isaac Gym, suggesting a focus on advanced simulation and AI-driven control for complex tasks. The robot’s tactile sensors and dexterous hands indicate a sophisticated AI stack for object manipulation and interaction.

Real-World Deployments or Pilots

 

While the snippets mention training for real-world roles and the GR-2’s capabilities for practical applications, the search results do not provide extensive details on current large-scale real-world deployments or pilots for the GR-2. The company’s background in rehabilitation robotics suggests potential applications in healthcare, but direct deployments for GR-2 are not explicitly stated.

Pricing (if known)

The provided search results do not explicitly state the pricing for the Fourier GR-2 humanoid robot.

Notable Achievements

 

  • Advanced Dexterity: The introduction of 12-DoF dexterous hands with tactile sensors marks a significant achievement, allowing for human-like flexibility and object interaction.
  • Robust Design: With 53 DoF, a height of 175 cm, and a weight of 63 kg, the GR-2 demonstrates a capable physical platform for humanoid robotics.
  • AI Training: Utilizing NVIDIA Isaac Gym for training highlights Fourier’s commitment to advanced AI and simulation for real-world applications.

 

 

 

 

Criticisms or Limitations

 

Based on the provided snippets, specific criticisms or limitations of the GR-2 are not explicitly mentioned. However, common challenges in humanoid robotics often include battery life, cost, and the complexity of real-world unstructured environments. The stated 2-hour battery runtime may limit continuous operation. The company’s plan for small-scale production might also indicate current manufacturing or market limitations.

Future Roadmap

 

Fourier Intelligence aims to continue expanding its robotics offerings, with a focus on general-purpose robotics. The company anticipates significant growth in the humanoid robot industry, predicting a tenfold increase. The GR-2 is positioned as a platform for practical applications, and ongoing training with tools like NVIDIA Isaac Gym suggests a roadmap towards more sophisticated AI-driven capabilities and broader real-world integration. The rebranding to Fourier and Fourier Rehab also indicates a strategic direction towards expanding its dedication to various applications.