NVIDIA introduces AI models and tools to advance autonomous vehicle technology ecosystem.

From NVIDIA: 2025-06-11 06:55:00

1. Autonomous vehicle (AV) technology is advancing with the development of unified, end-to-end architectures that rely on high-quality sensor data for training and validation.
2. NVIDIA has released Cosmos Predict-2, a new world foundation model that enhances future world state prediction capabilities for synthetic data generation, benefiting developers in the AV industry.
3. Cosmos Predict-2 accelerates AV training by improving context understanding from text and visual inputs, leading to better video generation with fewer errors and richer details.
4. Post-training Cosmos models unlock new training data sources by generating multi-view videos that enhance AV training datasets, improving model performance in challenging conditions like fog and rain.
5. AV companies like Plus and Oxa are leveraging Cosmos Predict to scale and accelerate vehicle development, utilizing synthetic data and multi-camera videos generated by the Cosmos platform.
6. In addition to Cosmos Predict-2, NVIDIA has introduced Cosmos Transfer as an NVIDIA NIM microservice preview, empowering developers with tools to augment datasets and generate photorealistic videos for AV development.
7. The integration of Cosmos Transfer and NuRec Fixer into the CARLA AV simulator enables developers to render synthetic simulation scenes with high fidelity and generate various lighting, weather, and terrain variations for testing and training.
8. NVIDIA Halos, a safety platform for AV systems, collaborates with automotive leaders like Bosch, Easyrain, and Nuro to ensure the safe integration of products with NVIDIA technologies and advance AV safety standards.

Read more at NVIDIA: NVIDIA Releases AI Models, Developer Tools to Advance AV Ecosystem