Fraunhofer Institute for Integrated Circuits IISThe development of quantum computers is progressing rapidly, but their use in industrial applications is still in its infancy. Yet quantum computers have the potential to fundamentally improve applications in many industries. The currently advancing increase in the capacity of quantum computers opens up the possibility of using quantum computers in AI systems. In AI, optimal control of dynamic systems is usually solved using reinforcement learning methods. First realizations of these methods with quantum computers already exist today.
QLindA – Quantum Reinforcement Learning
Goals and process
The project aims at combining these recent advances in the development of quantum computing with artificial intelligence, in particular for reinforcement learning (RL), and making it technically usable. To this end, the project investigates how RL can be implemented on quantum computers to solve a variety of relevant problems from industrial applications: RL-based control optimization in the process industry, the use of distributed automation systems in the smart factory, and optimization in production planning.
Innovation and perspectives
Due to the fact that quantum computers are error-prone, adaptations to classical algorithms must be made in order to ensure a certain degree of compatibility.
Contents of the project:
- Development of novel algorithms for quantum computers
- Benchmark for the evaluation of the methods
- Creation of a library to make them usable for industrial applications
- Investigation of possibilities and potentials as well as existing limitations of the algorithms
