IsaacLab

Isaac Lab is the GPU-accelerated, open-source framework for robot learning, unifying high-fidelity reinforcement learning (RL) and imitation learning (IL) workflows on NVIDIA Isaac Sim.

Isaac Lab is the official, GPU-accelerated, open-source framework for robot learning, built directly on NVIDIA Isaac Sim to leverage high-fidelity physics (PhysX) and photorealistic rendering (RTX). This unified, modular platform simplifies complex workflows like reinforcement learning (RL) and imitation learning (IL), crucial for robotics research. It comes 'batteries-included' with over 16 diverse robot assets (e.g., Unitree H1, Anybotics Anymal-C) and 30+ ready-to-train environments. The architecture enables rapid development and efficient sim-to-real transfer via features like domain randomization and vectorized rendering.

https://github.com/isaac-sim/IsaacLab