Requirements for the quality of synthetically generated sensor data are specified and metrics for the quality assessment and description of the synthetic sensor impressions are developed in order to achieve the most realistic simulation of sensor data possible.

The subproject investigates the required quality and coverage of data and how it can be verified that what has been learned using synthetic and augmented data can be transferred to reality.

Real data are recorded and methods are developed for the automated labeling of these data with annotations and context information as well as for the easy multiplication of the data by automated augmentation.

The cross-project provision of the developed tools and software platforms as well as the exchange on methodical-scientific procedures is prepared and the standardization of data formats and interfaces is initiated.