Hearken to this text |
Shanghai-based Fourier immediately launched GR-2, the most recent technology of its GRx humanoid robotic sequence. The corporate has upgraded its {hardware}, design, and software program.
“GR-2 is an enormous step into the way forward for humanoid robotics,” said Alex Gu, CEO of Fourier. “We’re enthusiastic about constructing probably the most intuitive embodied agent for AI, permitting it to have interaction with the bodily world in methods like by no means earlier than. Fourier is worked up to have builders, researchers, and enterprises be part of us on this unbelievable journey.”
This announcement adopted the firm‘s rebranding from Fourier Intelligence to Fourier earlier this 12 months, and the GR-2 launch builds on the manufacturing launch of the first-generation GR-1 in late 2023.
Fourier improves {hardware} design
The GR-2 stands 175 cm (68.9 in.) tall and weighs 63 kg (139 lb.), whereas the GR-1 is 165 cm (65 in.) tall and weighs 55 kg (121 lb.). GR-2 presents 53 levels of freedom and a single-arm load capability of three kg (6.6 lb.).
A brand new function of Fourier’s humanoid is a removable battery with twice the capability of its predecessor and a runtime of as much as two hours. For the reason that battery is now swappable, customers will be capable of swap batteries shortly and return GR-2 to work.
GR-2 options an built-in cabling design for energy and communication transmission, permitting hid wires and extra compact packaging. The environment friendly structure optimizes area for simpler modularization and larger adaptiveness for application-oriented customization.
To simplify the management system and scale back upkeep, Fourier redesigned GR-2’s joint configuration, shifting from a parallel to a serial construction. It mentioned this improves debugging, lowers manufacturing prices, and enhances the robotic’s potential to quickly be taught and transition from AI simulation to real-world purposes.
Dexterous palms have 12 levels of freedom
The robotic consists of palms with 12 levels of freedom, doubling the dexterity of earlier fashions. The palms are designed to reflect the flexibleness of human physiology and provide larger precision within the duties that the robotic might be requested to finish, mentioned Fourier.
The fingers on GR-2 even have six array-type tactile sensors that sense power and might determine object shapes and supplies. This allows new algorithms to optimize components dealing with, the firm mentioned.
Supporting a number of upper-limb instructing modes—digital actuality distant management, lead-through programming, and direct command—GR-2 can file a complete set of operational knowledge, from movement paths to tactile responses. Fourier mentioned it expects strong knowledge assortment to bridge the hole between digital fashions and real-world purposes, pushing the boundaries of robotic coaching and deployment additional.
FSA 2.0 powers dynamic mobility
To optimize its motion, Fourier developed seven distinct Fourier Sensible Actuators (FSA) for GR-2, every tailor-made to satisfy the precise torque calls for of every joint.
With peak torques exceeding 380 N.m (280.3 ft.-lb.), the FSA 2.0 actuators increase GR-2’s agility and dynamic capabilities. The twin-encoder system doubles management accuracy, guaranteeing exact actions even in high-pressure environments.
Designed for each velocity and precision, FSA 2.0 empowers GR-2 to navigate advanced duties with larger flexibility, mentioned Fourier.
Fourier optimizes instruments for open-source software program growth
Fourier optimized GR-2’s growth platform by introducing a brand new software program growth package (SDK) suitable with programming languages reminiscent of ROS. Builders can entry a set of pre-optimized modules for machine imaginative and prescient, path planning, and power suggestions management by means of utility programming interfaces (APIs), mentioned the firm.
Supporting frameworks reminiscent of NVIDIA Isaac Lab and Mujoco, the brand new platform empowers builders to give attention to innovation, streamlining their workflows, claimed Fourier.