Getting Started With Isaac Lab#

What You Will Build#

You will build a simulation workflow that uses an SO-101 arm to place a vial into a rack. You’ll first train a teacher policy using privileged information available only in simulation. Then you’ll use the teacher’s actions to train a student policy that uses a wrist camera and robot joint state—inputs available on a physical robot.

The interactive replay below may take a moment to load. It shows a simulation that you can view from different angles. You can also drag to rotate the view, scroll to zoom, and use the playback bar to pause or move through the recording.

Click here to open the replay in a separate window.

Choose Your Starting Point#

Isaac Lab helps you build robot learning workflows, including reinforcement learning and motion planning. This open-source framework connects robot learning tools with photorealistic rendering and GPU-accelerated simulation, so you can train many simulated robots in parallel.

In this learning path, you will learn:

Learning Objectives

  1. The concepts behind robot learning.

  2. How Isaac Lab supports robot learning workflows.

  3. How to create projects and tasks in Isaac Lab.

  4. How to train and evaluate policies.

You have a few options in how you can interact with this learning path (and the related learning paths). We recommend jumping to either Isaac Lab Basics or Train a Robot Arm and then following links back to fundamentals and theory as needed.