Safety Assurance of a Driverless Regional Train -Insight in the safe.trAIn Project

Marc Zeller, Thomas Waschulzik, Martin Rothfelder, Cornel Klein

Abstract: With the fully automated operation of trains a significant performance increase of railway systems can be achieved. However, the automation of the tasks of a human train driver, such as the track supervision, is only possible with the use of Artificial Intelligence (AI) and Machine Learning (ML). The problem, which still remains unresolved, is to find a practical way to link AI/ML techniques with the requirements and approval processes that are applied in the railway domain. The safe.trAIn project addresses this problem. Therefore, the project investigates methods to reliable design ML models, to prove the safety of AI-based functions taking robustness, uncertainty, and transparency aspects, and to setup an agile life-cycle for continuous engineering and assurance of ML-based system in railway.