Vehicle Data Management System for Scenario-Based Validation of Automated Driving Functions

Abstract

Proving the functionality of AI-controlled automated vehicles is a challenging task due to the enormous overall complexity. Although a scenario-based validation approach is widely accepted in the literature, the identification of these scenarios is still an open issue. Real-world test drives are valuable data sources for this purpose. However, an automated system is required for data management and scenario identification to analyze the vast amount of data in a legitimate amount of time and effort. Therefore, this work proposes a modular multi-tier Vehicle Data Management System for large-scale test campaign management and analysis as the basis for scenario-based validation of automated driving functions. For system demonstration, lane-change maneuvers are identified and extracted, and an onboard DAS is evaluated with a real-world test drive sequence.