Siemens: Digitalization and automation for driverless regional trains

In a project that will run until the end of 2024, Siemens and 16 partners will facilitate advances in the driverless operation of regional trains with the aid of artificial intelligence (AI). Within the “safe.trAIn” project, which the German government is subsidizing, there is a budget of €23 million available for this task. Solutions for meeting the requirements in this highly regulated and standardized environment have the potential to substantially boost the efficiency and sustainability of regional railway transportation.
Climate action and railway attractiveness
The transportation sector plays a crucial role when it comes to meeting targets for climate action. As part of the German federal government’s climate-action program, Germany is aiming to achieve – compared to the 1990 levels – a 40 percent to 42 percent reduction of CO2 emissions to 95 million to 98 million tons of CO2 by 2030. Reaching this goal will require making passenger-rail offerings more attractive. Building new railway lines – which requires some 20 to 30 years from planning to commissioning – takes an extremely long time and is very costly. As a result, the digitalization and automation of train operation within the existing rail network is a key lever for rapidly achieving successful outcomes. The improvements being targeted range from shorter headways – and thus more flexibility for passengers as a result of more frequent intervals between trains – to greater cost efficiency, and they extend all the way to a clear increase in the availability of rail transportation.
Artificial intelligence and functional safety in harmony
Based on the state of the art, conventional automation technology alone will not be enough to enable fully automatic railway operation. Artificial intelligence, however, offers major potential in this area. The challenge that has remained unresolved to date is that of finding a practicable way to link AI methodologies to the requirements and approval processes that apply in railway environments. This is where Germany’s government-subsidized safe.trAIn project comes into play. This project aims to lay a foundation for safe use of AI for driverless operation of rail vehicles and to thus address a key technological challenge hindering the adoption of driverless rail transportation.
Expanding driverless rail transportation
For several years now, solutions for completely driverless and unattended operation of trains have been successfully established on the market and in operations. Until now, however, these systems have been operating exclusively in controlled and closed environments, such as subway tunnels. Now, the safe.trAIn project is focusing on applying this technology for use in regional trains. Such trains operate in more open environments in which it is necessary, in particular, to reliably recognize obstructions – such as people on the lines as well as fallen trees or mudslides on the tracks, etc.
Project goals
The project goals are to perform integrated development of testing standards and of methods for using AI to automate rail transportation and to use example applications to verify the suitability of test standards. Focal points here will be on AI-based methods for driverless regional trains, approval-relevant validation of the product safety of the AI components, as well as testing processes and testing methods. Safe.trAIn will build on the results from the latest research and development activities and will continue the development of those activities in line with the new requirements. Important projects in this area are Shift2Rail, BerDiBa, ATO-Sense and ATO-Risk, and KI-Absicherung (“AI safeguarding”).
The participating project partners would like to use the project’s outcomes to launch automation solutions onto the market that enable highly automated and driverless operation of rail vehicles. In addition, relevant results from the safe.trAIn project are to be integrated into standardization activities in the areas of AI and rail transportation.