UT Austin Utilizes NVIDIA Isaac for Advanced Robot Training in Simulation
The University of Texas at Austin is leveraging NVIDIA Isaac's open robotics platform to train its humanoid soccer team, UT Austin Villa, and other robotics projects. This simulation-first approach allows students to develop and test robot skills in virtual 3D environments before deploying them on physical robots. The university has integrated eight Dell-built nodes, each equipped with eight NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, into the Texas Advanced Computing Center's (TACC) Stampede3 supercomputer. This infrastructure enables students to use NVIDIA Isaac Sim for creating simulated environments and NVIDIA Isaac Lab for training robot skills across numerous parallel simulations. This method addresses the challenge of limited physical robots by providing extensive hands-on experience for a larger number of students, allowing them to train complex behaviors like humanoid walking and high-speed off-road driving in a safe and scalable manner. The trained policies are then transferred to real rob...
