AI innovations for safe autonomous driving

    Politics and industry discuss the opportunities for the German automotive industry

    Press release

    Press release

    Berlin, October 06, 2024

    Project nxtAIM initiated — Generative AI to fundamentally change and accelerate the further development of autonomous driving — High-Level Event at the Federal Ministry for Economic Affairs and Climate Protection

    In line with the German automotive industry's safety requirements, the use of generative AI will enable the development of safe, highly automated driving functions in any traffic context. The next steps towards fully automated (Level 4) and ultimately driverless autonomous driving (Level 5) are now within reach.

    High-Level Event at the Federal Ministry for Economic Affairs and Climate Protection

    How this should happen in detail and why the time for it is now was the subject of an economic policy discussion panel today to which the Federal Ministry for Economic Affairs and Climate Protection (BMWK) invited. Federal Minister Robert Habeck discussed the innovative strength of Germany's leading industry in the promising areas of AI and autonomous driving with nine high-ranking figures from the automotive industry.

    The debate of about one hour involved development executives from the industry - Mercedes-Benz CTO Markus Schäfer, Continental Automotive CTO Gilles Mabire, Bosch Mobility CTO Dr. Mathias Pillin, President Valeo Germany Holger Schwab, AVL EVP Software Products, Emission and Services Jens Poggenburg and BIT TS managing director Cornelia Denk - VDA managing director Dr. Marcus Bollig and CEO Jülich Research Center Prof. Dr. Astrid Lambrecht. It focused on the opportunities for the German flagship industry, which is in the midst of the challenging transformation to electromobility and software-based vehicles.

    The reason for the meeting was the lighthouse project nxtAIM - Generative Methods for Perception, Prediction, and Planning, initiated by the VDA Lead Initiative for Autonomous and Connected Driving and funded by the BMWK. This project underlines Germany's leading position in the development and use of generative AI methods in the automotive context and at the same time represents a paradigm shift for the development of AI-based, autonomous driving functions. The project was made possible by the previous fundamental work in the four research projects of the AI family.

    Generative AI and autonomous driving: research project nxtAIM breaks new ground

    The ability of a vehicle to perceive its surroundings, predict how a traffic situation will develop and then navigate safely is based largely on machine learning and artificial intelligence methods. However, training autonomous driving functions requires not only large amounts of data, but above all driving data from traffic scenarios that are critical and safety-relevant. Such data collection has so far been associated with high costs and implies much time and effort. The result is now narrowly defined operating environments (Operational Design Domain (ODD)) that are associated with restricted vehicle speeds, weather conditions and times of day.

    The reasons for this are the linear system architectures and discriminatory methods of machine learning used today. Necessary development progress, particularly in terms of scalability (keyword: data, costs, computing capacity), transferability (keyword: expansion of operating environment (ODD)) and traceability (keyword: security and acceptance), is therefore only possible step by step. In the nxtAIM project, generative AI models form the research basis. Linear system architectures can finally be supplemented with the previously missing feedback paths in machine learning processes. In this way, foundation models for driving data are built that take into account the high requirements of safe autonomous driving.

    Provided there is enough computing power, generative AI methods can be scaled and trained relatively easily on huge data sets without a specific function reference. They can be used in the perception area, in the generation of sensor data, in predicting the behavior of road users and for route planning. In terms of function development, the industry is no longer limited by the amount of data actually collected, but almost unlimited new traffic and driving data can be generated synthetically. This opens up completely new possibilities for training, testing and validating driving functions, especially in rare and critical traffic situations, and allows a gradual expansion of the previously limited ODD.

    Three factors for project success

    Under the leadership of Continental and Mercedes-Benz, the project consortium will demonstrate over the next three years that successful European AI development in the field of autonomous driving is possible. Generative AI methods have reached the maturity required for the automotive industry. When people can no longer distinguish a generated image from a real photo, then the quality of the generated data is high enough to train and test perception systems. For the first time, the industrial partners involved in the project are making their driving data available - while complying with data protection regulations. With this data, which has been laboriously collected over the past few years, basic or foundation models can be developed that are suitable for the automotive sector. None of this would be possible without the extensive know-how of the project partners from industry and science involved. The collaboration with the Supercomputing Center at the Jülich Research Center in particular plays an important role in the project.

    Quotes

    Robert Habeck, Federal Minister for Economic Affairs and Climate Protection:
    "Germany already has a clear legal framework for autonomous driving. Now it is a matter of making highly automated and autonomous driving more widespread: step by step, always where it is already well suited, such as in public transport, and always with high safety requirements first. The use of artificial intelligence can spark a whole new dynamic here. Autonomous driving offers a wide range of economic opportunities for the German automotive and transport industry and for making mobility more sustainable in general. In order to further advance this highly relevant future topic with the safe use of artificial intelligence, we need a joint effort that focuses on the central core issues of the technologies. The AI models jointly developed in the nxtAIM research project based on the data treasures provided by the individual manufacturers and suppliers are a key way forward."

    Dr. Marcus Bollig, Managing Director of the German Association of the Automotive Industry VDA:
    "The nxtAIM project, funded by the BMWK, has transformed the national research ecosystem in autonomous and connected driving (AVF). Initiated and developed from the VDA flagship initiative on autonomous and connected driving, the entire industry benefits from the work of the partners in the initiative on pre-competitive AVF research questions. The project is the result of first-class cooperation between science, business and politics."

    Markus Schäfer, Management Board Member Mercedes-Benz Group AG, CTO Development and Procurement: "In 2021, Mercedes-Benz was the first manufacturer in the world to receive approval for highly automated driving according to SAE Level 3 here in Germany. At the end of September, we presented the fastest system for automated driving according to Level 3 for private cars to date: our DRIVE PILOT up to 95 km/h. At the same time, we all know that global competition is developing extremely dynamically. That is why we need an optimal interplay of research, cooperation and legal framework conditions in Germany and Europe so that we remain what we are: world leaders."

    Gilles Mabire, CTO Continental Automotive Technologies:
    "The mobility of the future needs new technologies and innovations, with new forms of cooperation. These include data and generative AI. Generative AI is the basic technology of the future. Together with partners from the industry and science, we are working on foundation models for the automotive sector that incorporate global knowledge about road traffic. Publicly funded research offers the opportunity to pool skills and knowledge for progress in AVF development."

    Astrid Lambrecht, CEO Jülich Research Center:
    "Large AI models, such as those required for autonomous driving, require top-class supercomputers on which they are trained. JUPITER, one of the most powerful AI computers in the world, is currently being created at the Jülich Research Center. It is being used jointly by science and industry in this project. The close collaboration on this unique machine is the key to innovations that will keep Europe sovereign and competitive."

    Mathias Pillin, CTO Bosch Mobility:
    "The issue of safety is fundamental to the breakthrough of autonomous driving. It is an enormous effort for the entire industry to ensure this safety even with automated driving functions. This is even more true when it comes to the use of AI, because here we need completely new procedures and security strategies. A practical proof of safety can only be developed across the industry."

    Jens Poggenburg, AVL EVP Software Products, Emission and Services:
    "The big challenge with autonomous driving is that in real operation, unplannable and unpredictable events in traffic must be safely resolved in real time. This is especially true in highly complex 'mixed traffic'. Integrated development along the entire value chain of the actors involved in autonomous driving is an important prerequisite for success. The use of generative AI in conjunction with virtualization can significantly increase the speed of development here and at the same time reduce costs."

    Holger Schwab, President Valeo Germany:
    "Research and development in the field of autonomous driving will remain essential in the future. We must continue to invest massively and continuously in order to maintain a strong market position internationally. Technology innovations such as generative AI, the willingness to cooperate pre-competitively and to define standards at an early stage enable the German and European automotive industry to achieve the innovation boost that is and remains important in international competition with the USA and China."

    Cornelia Denk, CEO BIT Technology Solutions GmbH: "Certifiable AI and a comprehensive, relevant database are of great value to the automotive industry as a whole. Innovative start-ups and specialized small and medium-sized technology companies are doing pioneering work for the entire industry and beyond. But politics should also initiate targeted funding programs that are specifically tailored to the needs of smaller companies."

    ©BMWK/Andreas Mertens
    Press Office

    Lena Anzenhofer

    Spokesperson with focus on security, digitalization, data, production and logistics