The new generation of general-purpose humanoid robot GR-2 independently developed by Fourier ushered in a blockbuster release. Based on the product vision of "creating the best embodiment carrier for AI", GR-2 has brought remarkable innovations and improvements in multiple key aspects such as hardware, design, and development framework, showing more flexible, powerful, and open features to meet the needs of developers in various fields and diverse application scenarios.
A complete upgrade in design, a new milestone for the GRx series
The GR-2 follows the bionic design language of the GR-1 in appearance, while further optimizing the full-body proportions - the upgraded height reaches 175cm, weighs 63kg, has a total of 53 degrees of freedom for the whole body, and the single-arm movement load reaches 3kg, which can complete more complex operations. The new generation battery capacity has been doubled, and the battery life has been extended to 2 hours. On the basis of supporting direct charging, a detachable battery swap scheme is added to fully meet the needs of sports.
The layout of the GR-2 adopts an internal cable design, which not only improves the overall aesthetics, but also simplifies the modular adaptation, reduces the obstacle of the developer's peripheral layout, and leaves enough space for scene adaptation and development. In order to optimize the control strategy and performance, the joint arrangement of GR-2 has been changed from parallel to series, so that the movement of each joint can be calculated independently, which is conducive to developers to deploy control algorithms, facilitate debugging and maintenance, and reduce manufacturing costs.
Humanoid tactile dexterous hands, more flexible upper limb operation
The GR-2 has significantly improved the flexibility and precision of upper limb manipulation, mainly due to Fourier's iterative upgrade of dexterous hand performance. The GR-2 is equipped with a new generation of dexterous hands, with 12 degrees of freedom in one hand, which perfectly reproduces the shape and size of the human hand, while simulating the natural movement of the human hand to the greatest extent.
Fourier has developed a new generation of dexterous hands
The new generation of dexterous hands is also equipped with 6 array tactile sensors and motion control algorithms, which can sense the grip strength in real time, adjust the gripping strategy, and achieve precise force control and grasping of objects of different shapes, materials and weights.
As the next main battlefield for the humanoid robot industry to actively promote commercialization, upper limb operation still needs to be trained by a large amount of data to achieve generalization and break the barriers of specificity ability. GR-2 supports a variety of upper limb teaching methods such as VR remote operation control, drag and drop teaching, and upper computer end control. Quickly create movements, record multi-dimensional data such as upper limb movement tracks, joint angles, strength, dexterous hand movements, and tactile information, providing strong support for collecting more real training data and helping developers bridge the gap between simulation and reality.
The GR-2 allows for more flexible upper limb maneuvering
Actuator 2.0, more powerful movement capacity
The GR-2 achieves a more anthropomorphic performance in terms of exterior structure and motion control. To this end, Fourier tailor-made 7 different models of FSA actuators, which are applied to the hip, ankle and other joints of the robot to meet the torque requirements of each joint. These self-developed actuators not only perfectly adapt to the bionic form of the robot, but also effectively reduce the hardware cost.
Fourier has developed its own new generation of FSA actuators
The new generation of actuators provides more powerful power for the GR-2, with a maximum joint peak torque of more than 380N.m. Combined with the perfect actuator control algorithm, the GR-2 has stronger load capacity and smooth motion control.
The new version of the actuator adopts a dual encoder system, which improves the encoder accuracy by 2 times, provides more accurate absolute position and speed feedback, and ensures the control accuracy of the GR-2 when performing various actions, and achieves a high dynamic response.
A more open development platform accelerates the implementation of AI embodiment
Fourier attaches great importance to the needs and experience of developers, and optimizes the original development platform in this product upgrade to form a set of development interface solutions based on mainstream programming languages. Whether you are a robotics researcher, an industrial automation engineer, or an AI software developer, you can quickly get started, efficiently customize a variety of applications, and realize robotics technology innovation.
The solution supports the algorithm program development of the server-client model, encapsulates a series of concise and easy-to-use APIs, and integrates pre-optimized algorithm modules such as machine vision, path planning, and force control feedback, which simplifies the implementation process of complex tasks while lowering the development threshold, and significantly improves the development efficiency. Currently, it supports open source frameworks such as NVIDIA Isaac Lab, ROS, Mujoco, and Webots.
The platform will provide detailed logging and remote monitoring functions, built-in strict security protocols and fault recovery mechanisms, ensure the safety and reliability of the development process, and provide developers with comprehensive and detailed documentation resources and rapid response technical support.
Fourier GRx, born for AI
In view of the future evolution of the humanoid robot ontology, Fourier proposed the concept of "product hexagon", focusing on six dimensions: motion intelligence, dexterous operation, cognitive intelligence, bionic design, user experience and commercial application, and clarifying the capability standards and development direction of the robot ontology.
"We've been thinking about the idea that there are so many of the world's strongest brains exploring AI, but it still needs a physical container to carry it," said Gu Jie, founder and CEO of Fourier. "When the humanoid robot ontology reaches the optimal level in all dimensions and becomes a hexagonal warrior, it will definitely be the best embodiment choice for AI. AI here is not limited to large models, but also includes the whole process of robot perception, decision-making, and execution. "
The Fourier GRx series is committed to creating more anthropomorphic, versatile, and friendly bipedal humanoid robots for scientific research institutions, leading enterprises, scenario developers, and end customers in its cooperative ecosystem, promoting the close integration of AI and the physical world, and realizing true human-machine symbiosis.