Greater safety for autonomous driving on roads and railways TU Braunschweig researches new safety concepts
In real traffic, things rarely go according to plan: a sudden rain shower, fog on the road or a branch on the tracks can all cause problems. In order for cars, buses, lorries and trains to be able to continue driving autonomously and safely in such situations in the future, they must learn to cope with the unexpected. This is precisely the goal of the new research project CONTROL, in which Technische Universität Braunschweig is involved. Industry and science are collaborating to develop new solutions that will ensure autonomous driving functions reliably, even in complex and changing environments, thereby enhancing safety, comfort and sustainability on roads and railways.
Autonomous driving is considered a key technology for future mobility. However, the automated systems already in use today only function under clearly defined environmental conditions. What is missing is the ability to operate safely in an open, highly dynamic environment with constantly changing conditions. CONTROL (short for “Controlling Risk of Highly Automated Transportation Systems Operating in Complex Open Environments”) addresses rare scenarios known as long-tail events, including unexpected obstacles on roads and railways, limited visibility and changing weather and light conditions. These rare events have a significant impact and are crucial for autonomous driving.
Two institutes from Technische Universität Braunschweig are involved in CONTROL. The Institute of Communications Engineering (IfN) is contributing its expertise in developing and evaluating artificial neural networks for perception through sensor technology in automated driving. The work focuses on the detection of sensor anomalies during environment perception runtime. Meanwhile, the Institute of Control Engineering (IfR) is focusing on verification and validation to support safety certification in uncertain conditions. The main objective of this work is to develop a scenario-based verification and validation concept that will enable the behavioural capabilities of an automated vehicle to be verified in defined scenarios. Additionally, the IfR is involved in developing the overall argumentation.
Over the next three years, the CONTROL project will develop innovative methods for validating autonomous vehicles within a dynamic road and rail system. A key element of this will be the development of cross-domain safety argumentation, which will systematically record, evaluate and control uncertainties in vehicle operation, integrating them into the architecture of autonomous systems. The system can then take cautious measures such as adjusting speed or performing a controlled evasion manoeuvre. This approach is supplemented by new quality metrics, adaptive sensor fusion and simulation-based validation.
The practical applicability of these methods will be tested in the test vehicles of the project’s partners throughout the project. Through its research, the project will strengthen the competitiveness of German industry in the field of autonomous mobility. It establishes a vital foundation for new products and standards, extending from road and rail to other application areas such as industrial automation and robotics.
About the CONTROL project
The pre-competitive research project, led by Siemens AG and Valeo Schalter und Sensoren GmbH, unites 24 industry and science partners, including OEMs, road and rail sector suppliers, and universities and research institutions. Supported by the VDA’s autonomous and connected driving initiative, the joint project is scheduled to run for 36 months and is funded by the Federal Ministry for Economic Affairs and Energy (BMWi).