Robot Simulation Software

Simulation is where robot policies are trained, benchmarked and stress-tested before touching real hardware, and increasingly a data source of its own through synthetic demonstration generation.

This list covers the simulators that matter for robot learning today, both corporate products and open-source projects.

Last updated 2026-08-09 · 8 companies listed

GPU-accelerated simulation and learning framework, the center of NVIDIA's robot learning stack.

MuJoCo (Google DeepMind)

United Kingdom / United States

Fast, contact-accurate physics engine, open-sourced by DeepMind; the default for control and RL research.

Long-standing open-source simulator tightly integrated with the ROS ecosystem.

General-purpose robot simulator widely used in education and research.

Open-source desktop simulator with a large library of ready-made robot models.

PyBullet

Open source

Python physics simulation built on Bullet; lightweight entry point for RL experiments.

Genesis

Open source

Newer open-source physics platform aiming at very high simulation throughput for robot learning.

Algoryx

Sweden

AGX Dynamics physics engine for heavy machinery and industrial simulation.

Need robot training data instead of a list?

AY-Robots collects demonstration data on real robot arms through a global teleoperator network, and sells ready-made datasets on the marketplace.