Am 16.11.2023 hält Dr. Cornel Klein von Siemens ab 16:00 Uhr einen Gastvortrag in der Ringvorlesung des Elitestudiengangs Software Engineering. Der Vortrag findet in Raum 1055N statt und hat den Titel "safe.trAIn – Engineering and Assurance of a Driverless Regional Train". Zum Vortrag sind alle herzlich eingeladen.

Abstract:

With the introduction of highly automated train operation (Grade of Automation (GoA) 4 operation) a significant performance increase of railway systems can be achieved. This includes the enhancement of the transport capacity in existing tracks, energy savings by means of an optimized driving strategy, reduced mechanical wear and tear as well as increased passenger comfort by means of homogeneous driving, and increased flexibility for demand-oriented train services. Traditional automation technologies alone are not sufficient to enable the fully automated operation of trains. However, Artificial Intelligence (AI) and Machine Learning (ML) offers great potential to realize the mandatory novel functions to replace the tasks of a human train driver, such as obstacle detection on the tracks. The problem, which still remains unresolved, is to find a practical way to link AI/ML methodologies with the requirements and approval processes that are applied in the railway domain. The safe.trAIn project (2022 - 2024) aims to lay the foundation for safe use of AI/ML for the driverless operation of rail vehicles and to thus addresses this key technological challenge hindering the adoption of unmanned rail transport. Within the project, which is funded by the German government, there is a budget of €23 million available for this task.

The project goals are to perform integrated development of guidelines and methods for the safety assurance of artificial intelligent in highly automated train operation. Based on the requirements for the certification process in the railway domain, safe.trAIn creates a safety argumentation for an AI-based obstacle detection function of a driverless regional train. Therefore, the project investigates methods to prove trustworthiness of AI-based functions taking robustness, performance, uncertainty, transparency, and out-of-distribution aspects of the AI/ML model into account. These methods are integrated into a comprehensive and continuous testing and verification process for trains. Moreover, a GoA 4 train architecture incorporating safe, AI-based functions for automated train operation is defined and assessed in terms of safety.  The feasibility of the guidelines and methods developed in safe.trAIn is evaluated with a real case study in which an exemplary safety case for a driverless regional train is created and assessed by auditors. Safe.trAIn builds on the results from the latest research and development activities (e.g., Shift2Rail, BerDiBa, ATO-Sense and ATO-Risk, and KI-Absicherung (“AI safeguarding”) and will continue the development of those activities in line with the new requirements.

The participating project partners are from the railway domain, academia, standardization bodies, and safety assessment bodies. The industrial partners will use the project’s outcomes to launch automation solutions in the market that enable highly automated and driverless operation of rail vehicles. In addition, relevant results from the safe.trAIn project will be integrated into standardization activities in the areas of safe and trustworthy AI and rail transportation.