Newton Fundamentals#

Newton Fundamentals is a guided, hands-on introduction to the Newton physics engine. We start with the high-level ideas, build practical intuition with small interactive simulations, and finish with a Franka cable pick-and-place task that combines inverse kinematics with Newton’s coupled-solver workflow.

Newton Physics Engine is an open-source, extensible physics engine built on NVIDIA Warp and OpenUSD, developed by NVIDIA, Google DeepMind, and Disney Research, and managed by the Linux Foundation

Because Newton is built on NVIDIA Warp, everything you build here runs on the GPU with near low-level performance while you stay in Python.

Understanding these fundamentals matters because they are the same abstractions Isaac Lab uses under the hood, which is exactly what we’ll explore in the next module.

How This Module Runs#

Choose the NVIDIA Brev or local setup in Getting Started: Newton Fundamentals.

This module should take about 2 hours to complete.

Learning Objectives#

By the end of this module, you’ll be able to:

  • Describe the Newton simulation abstractions and how a solver consumes them each step.

  • Build a scene with ModelBuilder, finalize it into a Model, and run a stepping loop.

  • Apply CUDA graph capture to reduce per-step launch overhead on the GPU.

  • Control a Franka arm with joint forces and joint position targets.

  • Solve batched inverse kinematics to follow a task-space path.

  • Assemble a coupled scene where a rigid arm manipulates a deformable cable.

What You’ll Build#

Each lesson introduces one idea at a time with a focused snippet, then hands you a single complete script you can run in isolation. The same scripts ship as standalone files in newton-fundamentals-examples. By the end you’ll have a small library of runnable Newton examples covering everything from a two-object drop test to a full coupled manipulation task.