Accelerating Robot Learning With Isaac Lab and Newton#
This course is a hands-on introduction to the Newton physics engine, and how it powers robot learning inside Isaac Lab.
We’ll start from the core Newton abstractions, build practical intuition with small simulations you run yourself. In the second course, we finish by training and playing back a reinforcement learning (RL) policy with Isaac Lab.
A preview of what you’ll build: a Franka arm driven through Newton-powered simulation, from basic control to a trained cube-lift policy.
We have both options to run this course on your own computer, or on the cloud with NVIDIA Brev.
Every concept is introduced with a focused code snippet, and at the end of each lesson we assemble those snippets into a single complete script to run on your own computer.
Who This Course Is For#
We designed this course for robotics engineers, simulation developers, and machine learning practitioners who have some background in robotics, simulation, or RL and want to add Newton to their toolkit. Comfort with Python is required. Prior exposure to NVIDIA Warp, MuJoCo, or GPU computing is helpful but not required.
What You’ll Learn#
By the end of this course, you’ll be able to:
Explain the core Newton abstractions:
ModelBuilder,Model,State,Control,Contacts, andSolver.Build and step a Newton simulation, and accelerate it with CUDA graph capture.
Control a Franka robot arm with joint forces, joint position targets, and batched inverse kinematics.
Assemble a coupled multi-solver scene for deformable cable manipulation.
Configure Newton as the physics engine inside Isaac Lab and train and play back an RL manipulation policy.
How the Course Is Organized#
Prepare your machine and launch the isolated course environment.
Core concepts, simulation loops, robot control, IK, and coupled manipulation.
Configure Newton in Isaac Lab and train a Franka cube-lift policy.