Discover the core aspects of the AI-Assurance-in-Mobility Europe 2025 conference. Experts in Berlin highlighted the crucial questions surrounding AI-supported driver assistance systems and autonomous driving at the intersection of technology and legislation.
The AI-Assurance-in-Mobility Europe conference celebrated a successful premiere at the beginning of April 2025 at the DIN-FORUM, Berlin. Numerous experts from politics and industry engaged in discussions about the technological and regulatory challenges in the field of Advanced Driver Assistance Systems (ADAS) and autonomous driving (AD), particularly with regard to the use of Artificial Intelligence (AI) to tackle the upcoming complex tasks. The following summarises the key aspects of the presentations given.
Global Harmonisation in Focus: The Path to Unified Rules for Autonomous Driving
A central point of discussion was the need for a global regulatory framework for highly and fully automated driving. Dr. Joachim Taiber, Founder and Managing Director of the International Alliance for Mobility Testing & Standardization (IAMTS), emphasised his organisation's efforts to create a globally accepted validation framework to accelerate the introduction of automated vehicles.
Richard Damm, President of the Federal Motor Transport Authority (KBA), provided an insightful overview of the development of such a global framework by the UNECE and its working groups. This included both new UN regulations for Driver Assistance Systems (DCAS) and the regulatory activities for Autonomous Driving Systems (ADS).
Advertisement
Type approval in practice: First experiences with highly automated vehicles
Another focus was on the practical experiences in the type approval of highly automated vehicles. Thomas Quernheim, Senior Vice President Engineering & Homologation Automotive TÜV Rheinland Group, shared insights into initial L3 and L4 projects and the associated challenges, particularly regarding the definition of operational design domains (ODD) and test scenarios. He pointed out that regulation is continuously evolving and process-based assessments allow manufacturers design freedom and faster updates.
AI as a key technology: Support for complex tasks in automated driving
The conference also addressed the significant role of AI in modern ADAS and ADS systems. Joachim Mathes, CTO Valeo Brain Division, presented the approach of end-to-end (E2E) automated parking, where deep neural networks are used for perception, prediction, and decision-making. He highlighted the advantages of holistic models but also emphasized the need for efficient and scalable learning through the use of large and unlabelled data sets as well as strategies like pre-training, fine-tuning, and distillation.
Statistical Safety versus Compliance: A New Perspective on the Validation of AI Systems
Dr. Daniel Fulger, Senior Manager Autonomous & Connected Mobility Capgemini Engineering, proposed the thesis that AI, from the perspective of complex systems, does not represent a fundamentally new technology but poses similar challenges in validation and testing as classical model predictive control systems (MPC). He argued for "statistical safety" and emphasised that the current perspective on validation and certification is not sustainable in the face of complexity.
Digital and Physical in Harmony: Guarantees for Product Safety of Automated Vehicles
The importance of digital and physical tests to ensure the product safety of automated vehicles was highlighted by Nils Katzorke, Project Manager for Test Technology Development at Mercedes-Benz AG. He demonstrated how test methods must be evaluated throughout the entire development process and that with increasing levels of automation, the requirements for test scalability grow. The use of digital twins for AV tests and the standardisation of credibility assessment can reduce development efforts.
Regulation of the Unknown: Technological Challenges in Autonomous Systems
Prof. Mario Trapp, Executive Director Fraunhofer-Institute for Cognitive Systems (IKS), highlighted the technological challenges in regulating the unknown in the context of automated driving systems. He emphasised that safety is not just a matter of assurance but a creative design task and advocated for simple, robust solutions as well as dealing with emergence in complex systems.
EU AI Act in Focus: Impact on the Automotive Industry
The impact of the EU AI Act on automated and autonomous vehicles was another important topic. Dr. Nils Lölfing, Council, Lawyer, Technology & Communications at Bird & Bird LLP, explained the fundamental requirements of the regulation and its potential implications for the automotive industry. He pointed out that although sector-specific laws take precedence, the AI Act will have an indirect influence on future delegated acts within the framework of type approval and a balance between innovation and safety must be found.
Automated Conformity Assessment: The EU AI Act in Practical Application
Dr. Benedikt Wolfers, Lawyer & Partner at Posser, Spieth, Wolfers & Partner, also addressed the interplay between EU legislation and the AI Act. He explained that future updates to the L4 Regulation 2022/1426 must take into account the requirements of the AI Act for high-risk AI systems. These include, among other things, that AI products must comply with all applicable EU laws, known and foreseeable risks must be analysed, and AI systems must consistently fulfil their intended purpose.
Sandy Rodrigues, Software Engineer AI & ML at Elektrobit GmbH, presented an approach for the automated compliance assessment of high-risk AI systems in accordance with the EU AI Act.
Generative AI for greater foresight: Next-generation object recognition and scene understanding
Dr. Christopher Plachetka, AI Engineer at Volkswagen ADMT GmbH, introduced the use of generative AI, particularly vision-language models (VLMs), for object recognition and scene understanding to improve the coverage of the operational design domain (ODD). He demonstrated how VLMs, trained on large internet datasets, can be used to extract scene and weather information.
Digital transformation of homologation: Increasing efficiency through AI and standardisation
Andreas Herzig, Partner Automotive at Deloitte, and Andreas Lauringer, CEO of Kontrol GmbH, presented digital solutions to increase efficiency and automate the homologation process through AI-supported tools and the standardisation of software quality characteristics. They emphasised the potential to significantly reduce testing costs by transitioning from physical to virtual tests and adopting a risk-based testing strategy.
EU Initiatives and Political Developments: Setting the Course for the Mobility of the Future
Dr. Stéphane Dreher, Head of Connected, Cooperative & Automated Mobility ERTICO - ITS Europe, provided an overview of EU legislative initiatives and political developments in the field of digitalisation and AI for the transport and mobility sector, including the AI Act, the Data Act, and the Chips Act. He highlighted the opportunities of AI to improve traffic management and autonomous technologies, but also pointed out the challenges of navigating regulatory frameworks and ensuring safety and data protection.
Conclusion: The Convergence of Technology and Regulation is of Central Importance
In conclusion, the AI-Assurance-in-Mobility Europe 2025 conference once again highlighted the central importance of the convergence of technology and regulation for the successful development and deployment of ADAS and ADS systems. The use of AI offers enormous potential to tackle complex tasks, but at the same time requires new approaches in validation, testing, and certification, as well as close cooperation between industry, politics, and standardisation organisations. The development of harmonised regulatory frameworks at a global level and the consideration of the specific challenges in the use of AI in safety-critical applications are decisive factors for the future of automated mobility.
The conference "AI-Assurance-in-Mobility Europe" is also set to take place in 2026. Further information about the event can be found here.