Nividia announced on Monday its Nvidia Isaac Sim on Omniverse is now available in open beta, giving roboticists essential tools to train robots for use in smart factories, warehouses and logistics systems.
Isaac Sim is a robotics simulation and synthetic data generation tool that allows robot makers to train and test robots more efficiently. Roboticists can provide more realistic simulations of a robot interacting with a virtual mockup of a real world in ways that Nvidia claimed can expand beyond what is possible in the real world.
The new release of Isaac Sim improves multi-camera support and sensor capabilities, with a CAD importer tool from PTC OnShape which makes it easier to test 3D objects. Such simulations can help in the design and development of a physical robot, then training of the robot and then deploying it as a digital twin in a photorealistic virtual world for testing.
More specifically, a new fisheye camera with synthetic data is supported along with an ultrasonic sensor, force sensor and custom lidar patterns. Robot Operating System 2 is supported.
A robot’s brain can be connected to the virtual world through the Nvidia Isaac SDK. https://developer.nvidia.com/isaac-sdk
In April at GTC Spring, Hai Loc Lu, lead system software engineer at Nvidia, provided an overview of new Isaac Sim features and included a chart that said Isaac Sim will be generally available about December.
A shortened version of his lecture is embedded here: