NVIDIA Isaac Sim & Omniverse
7 entriesGPU-accelerated, physically based simulation built on OpenUSD.
Gazebo & the ROS Ecosystem
5 entriesOpen-source robot simulation that works with ROS 2.
MuJoCo & Physics Engines
7 entriesFast, accurate dynamics for control and reinforcement learning research.
Other Open Simulators & Platforms
6 entriesAlternatives for manipulation, mobile robots, driving and game-engine twins.
Vendor Robot & Factory Simulation
6 entriesOffline programming and simulation tools from robot and industrial software vendors.
Digital Twin Platforms & Standards
10 entriesTwin platforms, modelling languages and interoperability standards.
Related Pages
5 entriesRelated directories on this site.
Frequently Asked Questions
What does simulation add to a digital twin?
A digital twin mirrors a real asset, and simulation lets you ask what would happen if conditions change. Physics-based simulators test robot programs, layouts and control policies virtually, so problems are found before they reach the factory floor or the road.
How do NVIDIA Isaac Sim and Gazebo differ?
Isaac Sim is built on NVIDIA Omniverse and OpenUSD and uses GPUs for physically based rendering, sensors and synthetic data. Gazebo is an open-source simulator with deep ROS integration that runs on modest hardware. Many teams use both, choosing by hardware, fidelity and licensing needs.
When is MuJoCo the right choice?
MuJoCo is popular for control and reinforcement-learning research that needs fast, accurate contact dynamics. It is open source under Apache 2.0, and MuJoCo Playground adds GPU-accelerated training environments. It is a physics engine rather than a full factory-twin platform.
Why use vendor tools such as ABB RobotStudio?
Vendor tools model that manufacturer's robots closely. ABB RobotStudio, for example, runs the same controller software as the real robot, which makes offline programming, reach checks and cycle-time estimates more trustworthy. The trade-off is that each tool mainly supports its own vendor's robots.
What is the sim-to-real gap?
It is the difference between simulated and real behaviour caused by imperfect physics, sensor noise and unmodelled effects. Teams narrow it with calibrated models, domain randomisation, synthetic data and validation on real hardware before deployment.