The NVIDIA Isaac GR00T Development Platform provides an open, modular, and fully integrated end-to-end workflow for developing, training, evaluating, and deploying humanoid robotics policies, spanning from simulation environment setup and teleoperation-based data collection to post-training and real-world deployment using the NVIDIA software stack.

GR00T 1.7 introduces a vision-language-action model pretrained on approximately 32,000 hours of real human demonstration and 8,000 hours of simulation, incorporating a Cosmos-Reason2-2B backbone (Qwen3-VL), robust ONNX/TensorRT export, and enhanced task decomposition for improved generalization, long-horizon reasoning, and cross-embodiment deployment.

Major robotics companies, research institutions, and XR device manufacturers have adopted components of the GR00T platform (Isaac Teleop, Lab, Sim, and ROS), leveraging its unified workflow to reduce integration complexity, accelerate skill development, and enable scalable AI-enabled humanoid deployments.

https://developer.nvidia.com/blog/develop-humanoid-robot-policies-end-to-end-with-nvidia-isaac-gr00t?ncid=so-nvsh-499222-vt48&es_id=0a01af6ec0