Below is a summary of the most common physics engines / tools for simulation:
PhysX is a real-time physics engine developed by NVIDIA. It focuses on providing high-performance, GPU-accelerated simulation of rigid bodies, fluid dynamics, cloth, and soft bodies. It’s widely adopted in games and simulations for its robustness, performance, and ease of integration with modern graphics pipelines. PhysX 4 is widely used for some years now while PhysX 5 is newer, contains more features and is more stable etc.
Used by: Unity, Unreal Engine, O3DE, Omniverse, and others.
Links:
Havok Physics is a commercial physics engine well-known for its stability, performance, and extensive toolset. It offers solutions for rigid body dynamics, character animation, vehicle physics, destructible environments, and more. Due to its mature, battle-tested technology, it has become a standard in numerous high-end video game productions.
Used by: Many AAA video games and engines, including titles like The Elder Scrolls (Skyrim), Halo, Assassin’s Creed series, and numerous other AAA franchises. It has also been integrated into custom engines used by large gaming studios and is sometimes licensed for industrial simulations. Integrations for Unreal Engine and Unity also exist.
Links:
Bullet is an open-source, cross-platform physics library known for its robust collision detection and soft body dynamics. It is widely used in games, visual effects, robotics, and VR due to its accessible open-source licensing, good performance, and active developer community.
Used by: Blender (for physics simulations within the 3D modeling and animation tool), Houdini (for rigid body solvers (RBD) simulations), and various robotics simulations (e.g., PyBullet for reinforcement learning and robotics research).
Links:
MuJoCo (Multi-Joint dynamics with Contact) is a physics engine designed primarily for research and development in robotics, biomechanics, and reinforcement learning. It focuses on accurate simulation of jointed bodies and advanced contact dynamics, providing a platform for studying control strategies and machine learning in physically realistic environments.
Used by: Robotics and AI research laboratories, OpenAI’s Gym environments (for training and benchmarking RL algorithms), DeepMind, and various academic institutions working on advanced locomotion and manipulation tasks.
Links:
Algoryx’s AGX Dynamics is a high-fidelity, multi-domain physics engine used in engineering simulation, training simulators, and virtual prototyping. It excels at simulating complex mechanical systems, heavy machinery, cables, ropes, and granular materials with a high degree of realism and industrial-grade performance.
Used by: Engineering companies, industrial simulation applications, maritime and construction equipment training simulators, VR-based design/testing platforms, and academic research institutions.
Links:
Project Chrono is an open-source multi-physics simulation engine that supports a range of applications from automotive and aerospace simulations to robotics, granular dynamics, and fluid-structure interaction. It emphasizes accuracy and flexibility, allowing coupling with external solvers and integration into various simulation workflows.
Used by: Academic research labs (for vehicle dynamics, biomechanics, and robotics), automotive and aerospace industries (for virtual prototyping and testing) and open-source projects requiring high-fidelity physics.
Links:
Altair offers a suite of CAE (Computer-Aided Engineering) tools including MotionSolve, a multi-body simulation software. It’s part of a larger ecosystem focusing on engineering design, optimization, structural analysis, and fluid dynamics. It’s often used for validating mechanical system behavior in the product design phase.
Used by: Automotive, aerospace, and mechanical engineering industries for virtual prototyping, performance optimization, and reliability studies. Companies use Altair’s solutions to design and refine complex machinery before physical prototyping.
Links: