Intrinsic and NVIDIA have collaborated to showcase advancements in robotic grasping and industrial scalability through the use of foundation models enabled by NVIDIA Isaac Manipulator. These models, based on a transformer deep learning architecture, allow for zero-shot learning and improved robot perception and decision-making. By using synthetic data generated on the NVIDIA Omniverse platform, Intrinsic was able to create a prototype for vacuum grasping in sheet metal and suction grippers.