NVIDIA Isaac Sim
Omniverse Robotics Experience: Isaac Sim 2020
leverages the powerful Omniverse Kit to build the next generation of robotics & AI simulator.
Realistic Physics
PhysX provides high-fidelity, stable, fast and GPU-enabled physics
RTX photo-realism
Real-time high-quality multi-GPU ray tracing and path tracing for high-fidelity simulation of complex 3D worlds
Connectivity
Easily connect the brain to a virtual world through Isaac SDK & ROS interface, fully-featured Python scripting, plugins for importing robot and environment models
Why Omniverse?
Omniverse Kit is NVIDIA’s state of the art platform for simulating the world; It natively supports RTX for photo-realism with real-time ray and path tracing, comes with built-in latest PhysX for stable, fast and GPU enabled physics, and supports MDL materials for physically based rendering.
Omniverse Kit provides extensible functionality and user-customized UI, python scripting for scene management, and supports workstation, headless and cloud deployment options.
Omniverse Nucleus enables seamless collaboration on building and sharing environment and robot models based on Universal Scene Description USD. Nucleus for Isaac Sim 2020.1 includes a starter set of robot and environment models.
Omniverse Connect facilitates robotics scene modeling through plugins for Maya, Revit, SketchUp, Rhino and Unreal Engine 4.
Isaac Sim: Omniverse Robotics Experience
Isaac Sim 2020.1 provides the essential features for building virtual robotic worlds and experiments. Isaac Sim supports navigation and manipulation applications through Isaac SDK and ROS, with RGB-D, Lidar and IMU sensors, Domain Randomization, ground truth labeling, segmentation, and bounding boxes.
Manipulation and Navigation with Isaac SDK
Carter navigates in Isaac Sim driven by Isaac SDK. A simple example with UR10 illustrates how a manipulator could be controlled from Isaac SDK.
Manipulation with UR10
In this example, a UR10 picks boxes from a conveyor and stacks them on top of each other on a pallet.
Manipulation with Franka
In this manipulation application Leonardo, a virtual reference robot for manipulation research based on Franka Emika Panda, stacks blocks and performs other tasks.
Domain Randomization
Domain randomization showing random changes in material (texture, color), light direction, light conditions, placement of objects to train and test robot perception.
Import Robot Models
Import your own URDF robot models, or build models directly from visual mesh using Omniverse tools.
Lidar and Cameras
Isaac Sim provides components for Lidar-type range sensing, RGB and depth cameras.
The Isaac Sim 2020.1 Early Access is offered both as a local desktop zip file and as a container that runs on Amazon Web Services (AWS) with streaming to the user’s desktop, to enable users who lack a workstation equipped with NVIDIA RTX.
Additional Resources
Ready to get started with Isaac Sim? Apply or log into the
Early Access Program
NVIDIA’s platforms and application frameworks enable developers to build a wide array of AI applications. Consider potential algorithmic bias when choosing or creating the models being deployed. Work with the model’s developer to ensure that it meets the requirements for the relevant industry and use case; that the necessary instruction and documentation are provided to understand error rates, confidence intervals, and results; and that the model is being used under the conditions and in the manner intended.