QLindA

Reinforcement Learning on Quantum Computers

In recent years, the latest advances in artificial intelligence (AI) have made it possible for AI systems to independently learn to play games such as chess or Go better than any human or computer before. The enabling key technology is called reinforcement learning and is now also used for adaptive control in industrial environments. The current rapid increase in the capacity of quantum computers opens up the possibility of using quantum computers in AI systems and offers the potential for ground-breaking increases in performance that could trigger a technological revolution with an impact on a wide range of applications.

Development of Innovative Algorithms

The project aims to combine the latest advances in quantum computing and artificial intelligence, especially in reinforcement learning, and make them technically usable. Based on existing scientific contributions, the project will investigate how reinforcement learning can be implemented on quantum computers in order to solve a variety of relevant problems from industrial applications: reinforcement learning-based control optimisation in the process industry, the use of distributed automation systems in the smart factory and optimisation in production planning.

The fundamentally different approach to classical algorithm design, which is coupled to the hardware, requires research into the transferability of classical approaches to quantum algorithms even before error-corrected quantum computers are available. The project will develop novel algorithms, create a benchmark for evaluating the methods and a library for utilisation in industrial applications, and investigate the possibilities and potential as well as existing limitations.