NVIDIA DRIVE Sim Powered by Omniverse

Accelerate autonomous vehicle (AV) development in the virtual world.

Simulation for Self-Driving Vehicles

Autonomous vehicles are one of the most complex engineering challenges of our time, requiring tools just as brilliant as the development teams. It isn’t possible to fully and safely test AVs in the real world, so NVIDIA DRIVE Sim™ provides a physically accurate simulation platform designed for fast and efficient AV testing and validation at scale.

Introducing the Neural Reconstruction Engine

The Neural Reconstruction Engine (NRE) is a set of AI tools that brings real world data directly into simulation, increasing realism and speeding up production. NRE turns video data collected during a drive into interactive 3D test environments where developers can modify scenarios, add synthetic objects, and apply randomizations to make the initial scene even more challenging.

DRIVE Sim for Closed-Loop Simulation

NVIDIA DRIVE Sim uses physically accurate simulation coupled with high-fidelity 3D environments to help create a virtual proving ground for AV development. Closed-loop testing can be done on individual software components or the entire AV stack. 

Expanding Capabilities with Omniverse

DRIVE Sim is built on the NVIDIA Omniverse™ platform, which provides the core simulation and rendering engines. NVIDIA Omniverse is architected with multi-GPU support to provide physically accurate, real-time multi-sensor simulation for autonomous vehicles with tight timing and accuracy requirements.

DRIVE Replicator for Synthetic Data Generation

With NVIDIA DRIVE Replicator™, developers can create diverse synthetic datasets for rare and difficult scenarios, including physically based sensor data and pixel-accurate ground truth labels. These labels include depth, velocity, occlusions and other hard-to-label parameters.

One Toolchain Across Development Teams (SIL and HIL)

DRIVE Sim is designed so all teams can use the same simulation toolset. It can run SIL on a local workstation or scale to multiple GPUs on one or more nodes. Hardware options are flexible–DRIVE Sim can run locally, on an OVX Server in the datacenter, or using a machine image from one of our many cloud partners. HIL or other XIL use cases can use NVIDIA DRIVE Constellation™ systems.

Simulate the User Experience

Designing for the user experience includes considering both the vehicle's behavior as well as interactions inside the vehicle. DRIVE Sim can be used for human in-the-loop testing and is the virtual studio where car designers and software engineers can collaborate and simulate the complete experience of the end user.

Open and Extensible

DRIVE Sim is open, modular, and extensible letting users customize the simulator to their needs. Extensions are easily built using the included SDK for sensor models, vehicle dynamics, traffic models, or interfaces to custom hardware. DRIVE Sim also has a rich ecosystem of partners who provide compatible extensions.

Meet a Few of Our Sim Ecosystem Partners

MathWorks logo

Rapidly import environments into DRIVE Sim with MathWorks RoadRunner.

Carla logo

Leverage a shared ecosystem of compatible simulation-ready content.

AVL logo

Quickly expand DRIVE Sim's V&V capabilities by connecting to Foretellix's coverage-driven validation platform Foretify™.

Aeva logo

Access FMCW lidar technology in simulation with Aeva’s sensor model in DRIVE Sim.

AEye logo

Access Arbe’s Perception Imaging Radar models in DRIVE Sim.

Cepton logo

Use DRIVE Sim in conjunction with Cepton’s high fidelity sensor models.

Hum3D logo

Bring assets directly into DRIVE Sim from Hum3D marketplace.

dSpace logo

Simulate the entire vehicle with dSPACE HIL and DRIVE Sim.

Luminar logo

Achieve detailed lidar simulation with Luminar’s sensor models in DRIVE Sim.

Read the Latest DRIVE Sim News

See All >

NVIDIA automotive solutions are available to automakers, tier 1 suppliers, startups, and research institutions shaping the future of transportation.