DFKI, Forschungszentrum Jülich, BMW Group, Mercedes-Benz AG, Volkswagen and Bosch want to jointly test AI applications for quantum computers. Quantum computers could take artificial intelligence (AI) and machine learning to a new level. However, the development of real AI applications for quantum computers is still in its infancy. The joint project Q (AI) 2, coordinated by Forschungszentrum Jülich, brings both approaches together on the basis of specific applications in the automotive industry. The three largest German car manufacturers BMW Group, Mercedes-Benz AG and Volkswagen as well as the supplier Bosch and the German Research Center for Artificial Intelligence (DFKI) are involved in the project.
Artificial intelligence is one of the most important key technologies in industry, especially in high-tech sectors such as the automotive industry. However, the computational effort for many applications is enormous. Even modern supercomputers sometimes need several days for certain tasks. Some problems have so far not even been solvable at all within realistic time spans.
“The use cases will, for example, be about optimizing flexible production processes in Industry 4.0, steering autonomous vehicles through traffic without collision, or intelligently planning electric bus tours. Working with specific tasks in the automotive industry as the start and end point of research is a key unique selling point of this project, ”explains Prof. Frank Wilhelm-Mauch from the Jülich Research Center.
The use of quantum computers and quantum annealers for real problems has so far hardly been researched due to the early development stage of these systems. Q (AI) 2 plays a pioneering role in this area. For the first time, researchers from academic institutions, automobile manufacturers and a supplier want to jointly create a broad base of quantum-accelerated AI algorithms that are optimized for the hardware available as well as for industrial issues.
In the sub-project QAICO of Q (AI) 2, the German Research Center for Artificial Intelligence (DFKI) is researching the possibilities and potential of quantum AI for complex AI problems of action coordination and cooperation. The application focuses on problems that are based on dynamic, flexible production flow planning in the context of Industry 4.0, collision-free navigation of self-driving cars and the formation of coalitions for optimal, cooperative service. In particular, it is investigated to what extent for which functional aspects and partial problem solutions of the selected, classical solution methods an adaptation for quantum computing by means of suitable existing and possibly further developed quantum algorithms in corresponding hybrid quantum AI processes is theoretically feasible and useful. In addition to a simulation on a classic computer, a possible execution of the prototypically implemented quantum AI processes on a real quantum computer is the subject of the investigations in QAICO. Processes and data are made accessible in an open software library of the overall project Q (AI) 2.
The desired findings could give German automobile manufacturers, for example, decisive competitive advantages: promising approaches should lead directly to concrete pre-development projects of the companies involved. At the same time, the results are made available to external users.
In addition, the partners in Q (AI) 2 want to determine meaningful key figures from which it can be seen when quantum computers can actually be used profitably for industrial applications. How many qubits and what cycle times must the systems have in order to achieve a real “quantum advantage”?
Project profile Q (AI) 2 https://www.quantentechnologien.de/forschung/foerderung/quanteninformatik-algorithmen-software-verbindungen/qlinda-1.html