tool· draft

NVIDIA Isaac Sim

NVIDIA's open-source robotics simulator built on Omniverse and OpenUSD; used to train, test and validate robots in physically based virtual environments and to generate synthetic training data.

Sensory landscape

Isaac Sim is NVIDIA's reference framework for simulating robots. It is built on Omniverse libraries and OpenUSD, imports robots and scenes from CAD, URDF and real-world captures, and is extensible enough that developers build their own OpenUSD-based simulators on it.

NVIDIA describes it as the place to train, test and validate a robot before it exists or before it is let loose: software-in-the-loop and hardware-in-the-loop testing, robot learning through the companion Isaac Lab, and synthetic data pipelines that can be augmented with Cosmos world models. It talks to ROS and ROS 2 through bridge extensions.

The code is on GitHub under Apache 2.0, subject to an additional NVIDIA software and materials licence, and is free to use; containers are on NGC and the AWS Marketplace, and cloud instances are offered through Brev, where the cloud provider's own fees apply. Vendors who redistribute Omniverse Kit with their product need an Omniverse Enterprise subscription. Status: operational.

form
open-source simulation framework
homepage
https://developer.nvidia.com/isaac/sim
licence
Apache-2.0, with an NVIDIA additional software and materials licence
cost
free

Tags: robotics, simulation, synthetic data, openusd, open source

staging: drafts and proposals shown