Industry News Oct. 09, 2024
Chinese robotics firm Fourier Intelligence has unveiled its new humanoid robot, GR-2, the latest addition to its GRx humanoid robot series.
The Shanghai-based firm was established in 2015 and has extensively used general-purpose robotics in medical and rehabilitation settings.
The GR-1, Fourier Intelligence’s first general-purpose bipedal robot, was introduced in mid-2023. This humanoid design was designed to help the elderly with various duties.
According to Fourier, it builds on direct customer feedback, and GR-2 features significant upgrades in hardware, design, and software.
“GR-2 is a big step into the future of humanoid robotics. We’re passionate about building the most intuitive embodied agent for AI, allowing it to engage with the physical world in ways like never before,” said Alex Gu, CEO of Fourier, in a statement.
Precision task handling
GR-2 builds on the success of its predecessor, GR-1, the world’s first mass-produced humanoid robot. Standing at 175 cm and weighing 63 kg, GR-2 offers 53 degrees of freedom and a 3 kg single-arm load capacity, allowing it to handle more complex tasks.
A key upgrade is its detachable battery, which has doubled its capacity. This extends runtime by an hour, enhancing its versatility for tasks like long-distance walking.
GR-2’s engineering advancements include integrated cabling for power and communication, optimizing space for easier modularization and greater adaptability. This compact design supports application-specific customization.
The humanoid’s 12-DoF hand is equipped with 6 array-type tactile sensors.
In addition, Fourier changed the robot’s joint structure from parallel to serial, streamlining the control mechanism, reducing maintenance expenses, and enhancing its capacity to go from AI simulations to practical applications.
GR-2 attributes 12-DoF dexterous hands that can mimic human flexibility for precise task handling while quadrupling the dexterity of earlier models. GR-2’s six array-type tactile sensors allow it to detect force, identify the materials and shapes of objects, and modify its grasp in real-time, allowing for the best possible manipulation in dynamic contexts.
It facilitates a range of upper-limb training modes, such as direct command, lead-through programming, and VR remote control, enabling thorough data capture of everything from motion trajectories to tactile feedback. By bridging virtual simulations with practical applications, this data collection advances the capabilities of robot deployment and training.
Innovative AI integration
Fourier aims to advance humanoid robotics with GR-2, incorporating human-like motion through seven custom-designed FSA actuators, each meeting the specific torque needs of its joints.
With peak torques exceeding 380 N.m, FSA 2.0 boosts GR-2’s agility and dynamic capabilities. The dual-encoder system enhances control accuracy, allowing precise movements even in demanding environments. Engineered for both speed and precision, FSA 2.0 enables GR-2 to perform complex tasks with greater flexibility.
Fourier is also concentrating on improving the GR-2 software development kit (SDK) developer experience. Compliant with popular programming languages such as ROS, the SDK provides pre-configured modules via user-friendly APIs for machine vision, path planning, and force feedback control.
The FSA 2.0 Actuators series includes seven different types of actuators, each tailored to meet the specific torque demands of GR-2’s joints.
The development platform enables developers to focus on innovation and streamline workflows by supporting frameworks like NVIDIA Isaac Lab and Mujoco. Developers now have cutting-edge resources at their disposal to create the next wave of humanoid robots thanks to GR-2.
Looking ahead, Fourier lists six essential areas for humanoid robotics development: locomotion, manipulation, cognition, bionic design, user experience, and economic feasibility.
According to a statement by the firm, GR-2’s launch marks a significant step toward future breakthroughs in human-robot collaboration and AI-driven advancements.
