Reconstruction of Scenarios for the Development of Automated Driving Functions

The article presents a holistic approach to reconstructing scenarios for the development of automated driving functions, addressing the bottleneck of testing and validation through the use of simulation. It covers the entire loop from raw measurements to the reconstruction and replay of driving scenarios in simulation, highlighting the importance of virtualization capabilities and AI tools. The authors describe two complementary approaches to generate scenarios: logical and concrete scenarios, each with its advantages and challenges. DeepScenario's AI Scenario Engine is introduced, which automates the reconstruction and extraction of driving scenarios using advanced 3D computer vision algorithms. The article also discusses the conversion of extracted scenarios into formats suitable for simulation platforms like CarMaker, enabling the testing of various environmental conditions and vehicle types. The benefits include decreased development time, increased robustness, and the potential for hyperscale simulation, making this approach a significant accelerator in bringing advanced driver assistance systems to market.