NVIDIA-Omniverse/PhysX

NVIDIA PhysX SDK

What it solves

PhysX provides a high-performance framework for simulating physical interactions in real-time, allowing developers to create realistic movement, collisions, and environmental effects like destruction and fluids.

How it works

The project consists of several specialized engines: the core PhysX SDK for general real-time physics, the Blast SDK for simulating fractures and destruction, and the Flow SDK for fluid and fire simulations. It also includes ovphysx, a C API with Python bindings that enables USD physics simulation and supports DLPack tensor interoperability for data exchange.

Who it’s for

Developers building real-time simulations, robotics researchers using tools like IsaacSim, and creators working within the NVIDIA Omniverse ecosystem.

Highlights

  • Real-time simulation: A dedicated SDK for high-performance physics.
  • Specialized effects: Separate libraries for destruction (Blast) and fluid/fire (Flow).
  • Python integration: Python bindings via ovphysx for easier scripting and USD physics simulation.
  • Tensor interop: Support for DLPack tensor interoperability to facilitate integration with AI and data workflows.

Related

  • Project
    bulletphysics/bullet3

    A professional C++ SDK for real-time collision detection and multi-physics simulation used in robotics, VR, and machine learning.

  • Project
    google-deepmind/mujoco

    MuJoCo is a general-purpose physics engine from Google DeepMind for simulating articulated structures that interact with their environment. It provides a high-performance C API, Python bindings, and interactive 3D visualization, making it a popular tool in robotics and machine learning research.

  • Project
    google/brax

    A fast, fully differentiable physics engine written in JAX for large-scale rigid body simulation and reinforcement learning training.

  • Project
    erwincoumans/tiny-differentiable-simulator

    A header-only C++ and CUDA physics library for rigid-body dynamics that supports automatic differentiation for optimization and learning in robotics.