Transforming your robotics projects
NVIDIA Isaac Sim and OpenUSD
NVIDIA Isaac Sim and OpenUSD
Scalable simulation technologies are fundamentally changing robotics development. They help to shorten development times and reduce costs - and at the same time create realistic test environments for AI-supported systems.
OpenUSD and NVIDIA Isaac Sim: tools for innovation
The Universal Scene Description (OpenUSD) provides a flexible platform for creating virtual worlds in which robots can develop their capabilities. On this basis, NVIDIA Isaac Sim, an application of the NVIDIA Omniverse platform, enables the simulation and testing of robots in physically precise environments.
New is the availability of Isaac Sim on Amazon EC2 G6e instances powered by NVIDIA L40S GPUs. These powerful instances not only make simulations more efficient, but also more accessible.
Translated with DeepL.com (free version)
Many companies rely on Isaac Sim to optimize robotic systems. Simulations can be used to run through complex scenarios that would be difficult or dangerous to implement in the real world. In addition, synthetic data allows AI models to be trained even when real data is limited or difficult to obtain.
A new workflow integration, based on OpenUSD NIM microservices, accelerates the creation of 3D datasets. Developers can train AI models for action recognition or other computer vision applications faster and more accurately.
The launch of NVIDIA Cosmos opens a new chapter for robots and autonomous systems. This platform combines generative modelling and video data processing to enable AI systems to better plan and execute actions.
Leading companies from the robotics and automotive industries, including Uber, Toyota and Agility, have already adopted Cosmos. The platform will be available as an open-source solution via GitHub, accelerating innovation in the industry.
Humanoid robots present an exciting challenge as they combine mechatronics, control and AI. With NVIDIA Isaac Lab, an open-source platform for robotics development, developers can train humanoid robots for demanding tasks.
NVIDIA's GR00T project develops new workflows for the simulation of humanoid robots and makes it possible to refine motion sequences and interactions in realistic environments. Systems such as MaskedMimic generate versatile movement patterns that can be applied to different terrains and scenarios.
Companies from a wide range of industries use Isaac Sim, OpenUSD and Omniverse to develop more powerful autonomous machines. Examples include:
Agility Robotics, which uses simulations to transfer robotic behavior directly into real robots.
Cobot, which simulates its collaborative robots to optimize logistics processes in warehouses and hospitals.
Wandelbots, which uses physics-based digital twins to test robots more easily and integrate them more quickly into real-life scenarios.
NVIDIA offers extensive resources to make working with Isaac Sim and OpenUSD easier. These include livestream sessions, technical training and free courses via the NVIDIA Deep Learning Institute (DLI) platform. The Alliance for OpenUSD also provides a central platform for sharing knowledge and best practices.
With the right tools and technologies, development processes in robotics can be made significantly more efficient. For more information about NVIDIA Isaac Sim and OpenUSD or about solutions that will advance your projects, please contact us at sysGen.
NVIDIA Isaac Sim and OpenUSD are revolutionizing robotics development with scalable simulation technologies. Organizations can efficiently train and test AI models in physically realistic virtual environments. Isaac Sim on NVIDIA L40S-based Amazon EC2 instances enables faster and more accurate simulations. Tools such as Isaac Lab and OpenUSD-based data management accelerate the development of humanoid robots and physical AI. With these technologies, costs can be reduced, time saved and realistic test scenarios safely implemented.