One of the main challenges of validating ADAS & AV systems is to analyze the data mountains of real drives to identify the gaps in the data.
When the gaps are identified, it is possible to generate synthetically the missing data in order to achieve full coverage for training and validation of ADAS & AV stacks.
In this session, we will explore how we ingest real data in order to find those gaps and how we parameterize our simulation platform to generate meaningful synthetic data for Perception, Prediction & Control stacks.