Using PhysX for Vehicle Simulations in Games

Driving is a part of everyday life. We know exactly how cars handle, so we are highly sensitive to the shortcomings of vehicle simulations in video games. The engineers behind NVIDIA PhysX 4.1 sought to address this by delivering a best-in-class vehicle simulation. They wound up building a solution that is so realistic, it’s now used as part of the foundation of NVIDIA’s self-driving car technology.

“The PhysX Vehicles extension library was written originally for games,” explains Kier Storey, Sr. Developer Technology Engineer at NVIDIA. “The entire purpose of it was to simplify adding vehicles to your games, so you didn’t need to have an expert in vehicle dynamics to be able to introduce driving scenarios. However, PhysX Vehicles is now being used in NVIDIA DRIVE SIM, which is our self-driving car training.”

In the video below, Kier details the accuracy of the vehicle physics model in PhysX 4.1.

This is an excerpt from a full GDC 2019 talk, PhysX 4: Raising the Fidelity and Performance of Physics Simulation in Games. That full video can be found here.
Comparing real vehicle performance data to the PhysX 4.1 simulation.

In the full presentation, Kier and Michelle Lu, Principal Software Engineer at NVIDIA, explain the latest features of PhysX 4, NVIDIA’s latest open-sourced PhysX version. Viewers will learn which new techniques are available, and how to use them. Examples are provided for how these techniques can be used to improve not only simulation quality, but also performance in a wide range of gaming applications.