NVIDIA Isaac Sim
NVIDIA Isaac Sim is an open source reference robotics simulation application built on NVIDIA Omniverse, enabling developers to design, simulate, test, and train AI-driven robots in physically realistic virtual environments. It is built atop Universal Scene Description (OpenUSD), offering full extensibility so developers can create custom simulators or seamlessly integrate Isaac Sim's capabilities into existing validation pipelines. The platform supports three essential workflows; large-scale synthetic data generation for training foundation models with photorealistic rendering and automatic ground truth labeling; software-in-the-loop testing, which connects actual robot software with simulated hardware to validate control and perception systems; and robot learning through NVIDIA’s Isaac Lab, which accelerates training of behaviors in simulation before real-world deployment. Isaac Sim delivers GPU-accelerated physics (via NVIDIA PhysX) and RTX-enabled sensor simulation.
Learn more
Parallel Domain Replica Sim
Parallel Domain Replica Sim enables the creation of high-fidelity, fully annotated, simulation-ready environments from users’ own captured data (photos, videos, scans). With PD Replica, you can generate near-pixel-perfect reconstructions of real-world scenes, transforming them into virtual environments that preserve visual detail and realism. PD Sim provides a Python API through which perception, machine learning, and autonomy teams can configure and run large-scale test scenarios and simulate sensor inputs (camera, lidar, radar, etc.) in either open- or closed-loop mode. These simulated sensor feeds come with full annotations, so developers can test their perception systems under a wide variety of conditions, lighting, weather, object configurations, and edge cases, without needing to collect real-world data for every scenario.
Learn more
Symage
Symage is a synthetic data platform that generates custom, photorealistic image datasets with automated pixel-perfect labeling to support training and improving AI and computer vision models; using physics-based rendering and simulation rather than generative AI, it produces high-fidelity synthetic images that mirror real-world conditions and handle diverse scenarios, lighting, camera angles, object motion, and edge cases with controlled precision, which helps eliminate data bias, reduce manual labeling, and dramatically cut data preparation time by up to 90%. Designed to give teams the right data for model training rather than relying on limited real datasets, Symage lets users tailor environments and variables to match specific use cases, ensuring datasets are balanced, scalable, and accurately labeled at every pixel. It is built on decades of expertise in robotics, AI, machine learning, and simulation, offering a way to overcome data scarcity and boost model accuracy.
Learn more
Gazebo
Gazebo is an open source robotics simulator that provides high-fidelity physics, rendering, and sensor models for developing and testing robot applications. It supports multiple physics engines, including ODE, Bullet, and Simbody, enabling accurate dynamics simulation. Gazebo offers advanced 3D graphics through rendering engines like OGRE v2, delivering realistic environments with high-quality lighting, shadows, and textures. It includes a wide array of sensors, such as laser range finders, 2D/3D cameras, IMUs, GPS, and more, with the ability to simulate sensor noise. Users can develop custom plugins for robot, sensor, and environment control, and interact with simulations via a plugin-based graphical interface powered by Gazebo GUI. Gazebo provides access to numerous robot models, including PR2, Pioneer2 DX, iRobot Create, and TurtleBot, and allows users to build new models using SDF.
Learn more