Automatic Large-Scale Synthetic Data Generation for Developing Fully Automated Trains

For the development, specification, and testing of fully-automated trains, access to vast amounts of sensor data covering the entire operational design domain (ODD) is critical — particularly for non-regular scenarios where real-world data is scarce or unavailable. By harnessing the power of NVIDIA Omniverse, we’ve developed a cutting-edge toolchain that traverses the ODD and orchestrates the simulation of both regular and non-regular railway scenarios at massive scale. This comprehensive synthetic data, combined with real-world data, creates a robust foundation for training advanced machine learning models specifically tailored for railway automation, overcoming the limitations of conventional data sources.