Studies of electrical engineering and information technology at the Karlsruhe Institute of Technology (KIT). Bachelor’s thesis at the Institute of Control Systems (IRS) on load forecasting in electricity distribution grids with high renewable energy penetration (2016). Following this, master’s studies with focus on control engineering and systems theory. Internship at the Autonomous Driving Campus of BMW in Unterschleißheim in the field of simulative validation of highly automated driving functions. Master’s thesis at the IRS on exploration strategies for cooperative controllers based on reinforcement learning methods (2019).
Since July 2019, member of the scientific staff of the Institute of Control Systems.
Modeling and Identification in Cooperative Human-Machine Scenarios
In almost all industrial sectors and even in the everyday life the amount of automated tasks increases. This leads to significant changes in the working and everyday environment. However, since the automation mainly concerns tasks and not complete jobs, humans more and more have to interact with highly automated systems. Examples can be found in the field of medical engineering, manufacturing engineering (fenceless factory) or automotive engineering.
In such situations, the human and the automation influence a technical system mutually and equally. In order to design the automation in such a cooperative human-machine scenario properly, a description resp. model of the human behavior is an important first step. Then, based on this model an identification algorithm will be needed to determine the crucial parameters, which describe how the human behaves. Designing the automation based on such an identified model provides the opportunity to get the most of the synergies of the cooperation.
Within the research project, describing the human as an optimal controller acts as a promising modeling approach. Thereby, the variability of human movements leads to additional challenges. In context of the research project, this variability is to be integrated and analyzed in the identification algorithms. A promising starting point is given by methods relating to Inverse Reinforcement Learning.