DB partners with NVIDIA to create a digital twin of railroads
Since 2019, Deutsche Bahn, as part of the Digitale Schiene Deutschland (DSD) initiative, has been implementing the Sensors4Rail project, which provides for the creation of an obstacle recognition system for unmanned trains. In such a system, artificial intelligence is used. For the training of AI it is necessary to reproduce many situations on the railway network under different weather conditions and at different times of the day. For this, real data obtained from on-board sensors or simulation modeling can be used. Railway Supply magazine writes about this with reference to zdmira.
To accelerate AI development and training, NVIDIA will provide dedicated NVIDIA DGX computing systems and NVIDIA OVX photorealistic modeling software, powered by the NVIDIA Omniverse real-time 3D modeling platform for creating and operating digital twins. DB and partners in the DSD initiative will take over the development of applications and the process of creating a digital twin of the railway network.
The basis of the digital twin is a detailed digital map that displays the geometry of the real line with high accuracy. When creating a digital twin, this map is augmented with 3D models, textures, and material performance data. As a result, photorealistic digital images of the line with tracks, stations, other infrastructure and surroundings are obtained.
To store and provide simulation data to users, there is a Data Factory, which DB started to build in the first half of 2022. German Railways expects the next phase of this data factory to be implemented at the European level to cover international rail traffic. Relevant research is being conducted within the Europe’s Rail partnership.
Railway news you might have missed:Facebook, Twitter, LinkedIn, read Railway Supply magazine online.
Place your ads on webportal and in Railway Supply magazine. Detailed information is in Railway Supply media kit