Voxel51, Nebius, and NVIDIA Power Porsche's Synthetic AV Data Pipeline

Porsche Research, in collaboration with Voxel51, NVIDIA, and Nebius, is advancing its synthetic data generation pipeline to enhance autonomous vehicle training by addressing the limitations of real-world data collection, which is often slow, costly, and insufficient for capturing rare, high-stakes scenarios. The NVIDIA Physical AI Data Factory Blueprint serves as an open reference architecture, enabling seamless data augmentation and analysis workflows through modular processes that transform raw driving footage into diverse, high-quality training sets. Voxel51's data curation tools and Nebius's GPU cloud infrastructure streamline the generation and evaluation of synthetic data, ensuring model-ready outputs that pass rigorous quality checks. This collaboration empowers Porsche to close gaps in long-tail distribution by automating the identification and augmentation of impactful scenes, thereby reducing pipeline complexity and paving the way for agentic workflows that minimize human intervention.