NVIDIA DRIVE Sim
Type: Platform Tags: NVIDIA, DRIVE Sim, autonomous vehicles, simulation, synthetic data, Cosmos, Omniverse, NuRec, AV Related: NVIDIA-Drive-Platform, NVIDIA-Cosmos, NVIDIA-Omniverse, NVIDIA-Cosmos-Curator-LHA, NVIDIA-NIM, NVIDIA-DRIVE-AGX-Thor Sources: https://developer.nvidia.com/drive/drive-sim, https://developer.nvidia.com/drive Last Updated: 2026-04-29
Summary
NVIDIA DRIVE Sim is NVIDIA’s autonomous-vehicle simulation and synthetic-data development surface. Current NVIDIA DRIVE Sim public material emphasizes AV simulation pipelines powered by NVIDIA Cosmos world foundation models, NVIDIA NuRec neural reconstruction, Cosmos Transfer/Predict, physical AI open datasets, Cosmos Curator, and dataset search. It connects DRIVE development to NVIDIA-Omniverse, NVIDIA-Cosmos, and NIM-style model services for generating and validating AV data before deployment on DRIVE AGX platforms.
Detail
Purpose
Autonomous vehicle teams need large-scale scenario coverage, synthetic data generation, reconstruction of real-world environments, and closed-loop validation that cannot be achieved safely or economically through road testing alone. DRIVE Sim provides a simulation entry point for AV workflows using NVIDIA physical AI, world model, and Omniverse technologies.
Key capabilities
- AV simulation and synthetic data generation workflows for training, testing, and validation.
- NVIDIA Cosmos world foundation models for post-training, scenario generation, and video/world generation.
- NVIDIA NuRec neural reconstruction for reconstructing real-world environments into simulation-ready assets.
- Cosmos Transfer and Cosmos Predict workflows for AV data generation.
- Cosmos Curator and dataset search for filtering, annotating, searching, and retrieving AV datasets.
- Open-source AV simulator integration signals, including CARLA integration with NuRec and Cosmos Transfer.
- Physical AI open datasets and starter scene packs for AV workflows.
NVIDIA context
DRIVE Sim is no longer just a standalone simulator in the wiki sense; current public NVIDIA messaging makes it a bridge between NVIDIA-Drive-Platform, NVIDIA-Cosmos, NVIDIA-Omniverse, and NVIDIA model services. It should stay as one DRIVE simulation page, with NuRec and Cosmos sub-workflows folded here unless a separate durable NVIDIA NuRec docs surface becomes necessary.
Connections
- NVIDIA-Drive-Platform - parent AV platform that uses simulation for data generation, testing, and validation.
- NVIDIA-Cosmos - world foundation model platform used for AV world/video generation and synthetic data.
- NVIDIA-Omniverse - simulation and digital twin substrate behind NVIDIA AV simulation workflows.
- NVIDIA-Cosmos-Curator-LHA - adjacent video/data curation workflow for large physical AI datasets.
- NVIDIA-NIM - current DRIVE Sim material references NVIDIA NIM microservices for model/API workflows.
- NVIDIA-DRIVE-AGX-Thor - target DRIVE hardware for validating and deploying AV workloads after simulation.
Source Excerpts
- NVIDIA DRIVE Sim public pages describe AV simulation and SDG pipelines powered by NVIDIA models, tools, and libraries.
- The page highlights Cosmos world foundation models, NuRec neural reconstruction, Cosmos Transfer/Predict, and physical AI open datasets.
- NVIDIA positions Cosmos Curator and dataset search as tools for accelerating AV data processing and retrieval.