Intelligent Data Recording in the Automotive Field (IDA)
The validation of autonomous vehicles is one of the key aspects in current research in the automotive field. Beside the confidence of the public and the customers, it is also of legal relevance. A statistical significance demonstration that an autonomous vehicle causes 20% less fatalities than the average German driver would require more than 10 billion kilometers to be driven (Fig. 1). Driving such a distance is economically hardly possible. A commonly used alternative is the scenario-based approach, where only representative scenarios are tested for validation. The identification of representative, relevant and important scenarios contributes to the reduction of the testing effort. Moreover, such a detection reduces the required storage for recorded data. The objective of the project is to identify novel and relevant scenarios, using rule-based methods and machine learning.