Over the years OEMs and Tier1s for sensor-based systems in the area of driver assistance and autonomous driving focused on harvesting as much data as possible from real world test drives for validation. As petabytes accumulated from sensor raw data now a paradigm shift happens. Since there shouldn’t be more duplicate scenes in the database, now only the relevant corner cases for algorithm developers should be recorded. This technical presentation shows a holistic measurement architecture, which allows an intelligent recording. It covers filtering of relevant scenes in the vehicle and managing the test fleet efficiently. Engineers get a technical overview of a modular measurement architecture which covers the needs for processing the high data rates in the vehicle for sensorics like Camera, Radar and LIDAR Algorithm developers can see how they filter corner cases already in the vehicle Technical management gets a recommendation how to increase efficiency in test fleets