GR00T
A foundation model released by NVIDIA for embodied intelligence, designed to provide humanoid robots with general-purpose perception and motion pretraining capabilities. It adopts a pretraining-then-fine-tuning paradigm analogous to that used in large language models.
Timeline (last 90 days)
Typical representatives of VLA models include RT-2, OpenVLA and GR00T, which are usually pretrained jointly on internet-scale image-text data and large-scale robotic manipulation data.
In the field of robot foundation models, NVIDIA's GR00T series already exists, and in the open-source community, VLA models such as RT-2 and OpenVLA are iterating rapidly.
NVIDIA's GR00T emphasizes the combination of "foundation model + simulation data + hardware platform," relying on the Isaac simulation environment to generate large-scale synthetic trajectories.
LIBERO simulation successfully runs long-horizon tasks on Mac for the first time, with GR00T achieving a 100% success rate, WAM achieving 98%, and a 2.8-billion-parameter VLA small model achieving 96%.
NVIDIA released GR00T, a foundation model for embodied intelligence, aimed at providing humanoid robots with general-purpose perception-motor pretraining capabilities.
GR00T adopts an approach aligned with the large language model pretraining paradigm, first establishing general representations on massive multimodal data and then fine-tuning for specific tasks.
All Facts (6)
Typical representatives of VLA models include RT-2, OpenVLA and GR00T, which are usually pretrained jointly on internet-scale image-text data and large-scale robotic manipulation data.
50%UnverifiedIn the field of robot foundation models, NVIDIA's GR00T series already exists, and in the open-source community, VLA models such as RT-2 and OpenVLA are iterating rapidly.
50%UnverifiedNVIDIA's GR00T emphasizes the combination of "foundation model + simulation data + hardware platform," relying on the Isaac simulation environment to generate large-scale synthetic trajectories.
50%UnverifiedLIBERO simulation successfully runs long-horizon tasks on Mac for the first time, with GR00T achieving a 100% success rate, WAM achieving 98%, and a 2.8-billion-parameter VLA small model achieving 96%.
50%UnverifiedGR00T adopts an approach aligned with the large language model pretraining paradigm, first establishing general representations on massive multimodal data and then fine-tuning for specific tasks.
50%UnverifiedNVIDIA released GR00T, a foundation model for embodied intelligence, aimed at providing humanoid robots with general-purpose perception-motor pretraining capabilities.
50%