From ODDs to Simulation: Automatic Railway Scenario Generation

The safe deployment of autonomous railway systems requires systematic testing across a wide range of representative and safety-critical scenarios. Defining these scenarios in a structured and reproducible manner is a key challenge, as they must accurately reflect the system's intended Operational Design Domain. For the railway domain, however, systematic approaches for defining and generating such scenarios remain limited. This paper presents a methodology for automatic scenario generation that bridges the gap between formal Operational Design Domain definitions and simulation-based testing of autonomous railway systems. We propose a structured framework to derive simulation scenarios from Operational Design Domain specifications by identifying relevant operational parameters, environmental factors, and interaction types. In this context, an Operational Design Domain for railways is defined as the set of environmental, infrastructural, and operational conditions under which an autonomous train system is designed to operate safely. The approach formalizes Operational Design Domain components, such as infrastructure, operational constraints, and external agents, and defines how they can be parameterized and combined to form meaningful scenario classes. This structured representation ensures that the generated scenarios are traceable to system-level requirements and representative of real operational contexts. Building upon this methodological foundation, we demonstrate its implementation within the CARLA simulation environment. Although originally designed for road-based traffic, CARLA's open and modular architecture enables its extension to railway applications. We introduce adaptations that support track-based motion, signaling systems, station layouts, and multi-modal interactions, as well as the automated generation of OpenSCENARIO templates aligned with the defined Operational Design Domain structure. The resulting framework enables scalable and reproducible scenario generation directly linked to Operational Design Domain specifications, supporting systematic validation and verification processes for autonomous railway technologies. By connecting conceptual Operational Design Domain definitions with executable simulations, the approach facilitates structured scenario-based testing and provides a foundation for simulation-driven safety assessment and reliability evaluation in the development of future autonomous rail systems.

Keywords: Operational Design Domain (ODD), Scenario generation, OpenSCENARIO, CARLA.