Skip to main content
Gizmo is built around the two simulators most widely used in robotics research and development: NVIDIA Isaac Sim and MuJoCo. Every asset and scene you create in Gizmo can be exported directly to either simulator in its native format — no manual conversion, no plugin installation, no hand-editing of config files.

​
NVIDIA Isaac Sim

NVIDIA Isaac Sim is a GPU-accelerated robotics simulation platform built on the Omniverse USD (Universal Scene Description) framework. It is widely used for photo-realistic rendering, synthetic data generation, and large-scale parallel simulation of robot fleets. Export formats: USD (.usd / .usda) and URDF (.urdf)
  • USD is the native Omniverse format. It carries the full scene graph — meshes, materials, physics articulation trees, and joint definitions — in a single portable file.
  • URDF (Unified Robot Description Format) is an XML-based format used across the ROS ecosystem and supported natively by Isaac Sim’s URDF importer.
When you export from Gizmo to Isaac Sim:
  1. Open Isaac Sim and use File → Import (USD) or the URDF Importer extension.
  2. Select the exported file from Gizmo.
  3. The asset or scene loads with physics properties, articulation, and materials intact.
Isaac Sim’s GPU-accelerated physics backend (PhysX) supports large numbers of simultaneous simulation instances — useful for reinforcement learning and synthetic data pipelines.

​
MuJoCo

MuJoCo (Multi-Joint dynamics with Contact) is a fast, lightweight physics engine developed by DeepMind. It is the de facto standard for robotics control research, especially reinforcement learning, due to its speed, accuracy, and small computational footprint. Export format: MJCF (.xml) MJCF (MuJoCo Modeling Format) is an XML dialect that describes bodies, joints, geoms (collision and visual geometry), actuators, and sensors in a single file. Gizmo’s MJCF export includes:
  • Bodies and joints — the full articulation tree with joint types, axes, and limits
  • Geoms — both visual and collision geometry meshes referenced from the XML
  • Inertial properties — mass and inertia tensors per body, derived automatically during generation
When you export from Gizmo to MuJoCo:
  1. Download the MJCF export (.xml file plus associated mesh assets).
  2. Load with mujoco.MjModel.from_xml_path("your_asset.xml") in Python, or open directly in the MuJoCo viewer.
  3. Run your simulation — physics are active immediately.
MuJoCo’s constraint solver is highly accurate for contact-rich manipulation tasks, making it ideal for grasping, dexterous hands, and legged locomotion research.

​
Choosing Your Simulator

Best for:
  • Photo-realistic rendering and synthetic data generation
  • GPU-accelerated parallel simulation (many robot instances at once)
  • ROS 2 integration and sensor simulation (cameras, LiDAR, IMUs)
  • Large warehouse or factory-floor environments
Export format: USD (.usd) or URDF (.urdf)Hardware requirement: NVIDIA GPU (RTX recommended)Physics backend: PhysX (NVIDIA)Typical use cases: sim-to-real transfer, synthetic dataset creation, fleet simulation, perception pipeline development
Isaac SimMuJoCo
Rendering qualityPhoto-realistic (Omniverse)Basic (viewer)
Simulation speedGPU-acceleratedVery fast on CPU
Primary formatUSD / URDFMJCF (XML)
ROS integrationNative ROS 2 bridgeVia external wrappers
Hardware neededNVIDIA GPUAny CPU
Best workflowSynthetic data, fleet simRL training, manipulation
Support for additional simulators is on the roadmap. If you need a specific format — such as Gazebo SDF, Webots, or PyBullet URDF — reach out at viswajit@antimlabs.com to share your requirements.
For hands-on export instructions, see the Isaac Sim guide and the MuJoCo guide.