Machine Learning in Optimal Control for Industry

We specialize in integrating Artificial Intelligence with physical insights to tackle real-world industrial challenges. Our dedicated team with a strong background in engineering, physics and mathematics harnesses the power of advanced algorithms, machine learning, and physics-based models to drive innovation and efficiency.

What sets us apart:
We combine a strong background in physics, engineering and mathematical optimization with a variety of evolutionary and Artificial Intelligence methods. Physics and engineering allow us to include domain knowledge and take advantage of explainable first principle models that can be implemented from comparably little data. On the other hand, in a data-driven method, an Artificial Intelligence system learns a model from a large collection of data. This reduces the cost of modeling and ensures flexibility and continuous adaptability under varying external conditions. We are convinced that a customized combination of both approaches is the key to success in many cases.

We are used to the challenges of working with real-life environments - as opposed to laboratory environments, where collecting data and running dedicated experiments is often challenging. We specialize in optimizing complex processes with regard to cost and resource efficiency. Our portfolio of projects includes optimizations of industrial sieve systems and a cement plant as well as heating systems for small residential buildings. Our toolbox includes a variety of methods from classical control and optimization to evolutionary and machine learning approaches.

We are a subgroup of the Applied Complex Systems Science research area of the Institute of Applied Mathematics and Physics.