Impulse Project 13: Hybrid Models and AI Methods for Safe Mobility (HyMne 2) - Data Generation and Data Quality

IP13 is divided into three subprojects. Subproject 1 focuses on the task of highly accurate state estimation using inertial measurement systems. In this subproject, correction signals are calculated from onboard sensors, e. g. speed or slip angle signals. The estimation methods employed use a combination of AI methods and physical models. Subproject 2 uses the results from IP8, where it was shown that the combination of vehicle motion models with deep learning architectures is highly beneficial for behavior prediction and generation of drivable trajectories for road users. The developed hybrid models are extended in IP13 by introducing guidance aspects such as goal points, traffic signals and acceleration profiles. This increases the relevance of the hybrid models for simulation environments. In subproject 3, the focus is on the identification and selection of traffic scenarios that are critical or exhibit atypical driver behavior, so-called "corner cases". These scenarios are selected from publicly available data sets as well as from traffic scenarios generated with the methods researched in IP8.