While massive datasets like Waymo or nuScenes capture extensive real-world behavior, raw data alone isn't a validation asset. Using the NVIDIA Physical AI Autonomous Vehicles dataset as a case study, this presentation introduces a transferable workflow for converting proprietary fleet data into structured, reusable scenarios. We demonstrate a full pipeline: extracting trajectories, classifying events, and mapping data to OSC2 scenario primitives. By quantifying coverage against a defined Operational Design Domain (ODD) and performing gap analysis, we show how to transform raw recordings into actionable insights that feed directly into verification plans and support critical ADAS/AD release decisions.

• Transform Raw Data into Assets: Learn to convert massive volumes of raw fleet recordings into structured, reusable verification and validation assets.
• Standardized Scenario Mapping: Master the workflow for extracting vehicle trajectories and mapping events to OpenSCENARIO 2.0 (OSC2) primitives.
• ODD Coverage Quantification: Discover how to quantify data coverage against a defined Operational Design Domain to identify critical testing gaps.
• Transferable Data Workflows: Apply a proven case study approach to your proprietary data for consistent event recognition and classification.
• Data-Driven Release Decisions: See how coverage gap analysis feeds directly into validation plans to support confident ADAS/AD release decisions.