Institute of Control Systems (IRS)

M. Sc. Florian Köpf

  • Karlsruher Institut für Technologie (KIT)
    Campus Süd
    Institut für Regelungs- und Steuerungssysteme
    Geb. 11.20 (Engler-Villa)
    Kaiserstr. 12
    D-76131 Karlsruhe

Curriculum Vitae

Studies of electrical engineering and information technology at the Karlsruhe Institute of Technology (KIT). Bachelor’s thesis at the Institute of Industrial Information Technology (IIIT) on the fusion of depth and RGB data for pedestrian detection (2013). Internship at the Automotive Products Research Laboratory of Hitachi America Ltd. in Detroit. Master’s thesis at the Institute of Control Systems (IRS) on the development of an inverse reinforcement learning method for modeling human movement behavior (2016).

Since June 2016, member of the scientific staff of IRS.


Design of cooperative control concepts taking into account insecure information

Intelligent cooperation of humans and machines leads to synergies when the strengths of both partners are combined. While man has strong associative and cognitive capabilities as well as creativity and flexibility, automation is characterized by high reliability and precision as well as lacking fatigue. Cooperative control systems, in which man and machine simultaneously interfere with dynamic systems, enable safe and efficient control of of these systems in a number of domains, such as driver assistance systems, industrial production plants, robotics, medical engineering and rehabilitation as well as aerospace technology.

An important aspect of this concept is the question how both partners select their control actions while simultaneously acting on a dynamic system. This decision making is based on models and information on the state of the system and the objectives and capabilities of the partners. The underlying models and data are usually imprecise and incomplete. The research project is aimed at developing cooperative control concepts taking into account uncertainties and lacking knowledge in decision making. The resulting challenge is how controllers can be designed to provide best possible support in reaching the objectives of man and machine in spite of imprecise or lacking information on the behavior of both man and dynamic system. A potential approach is based on probabilistic conclusion processes, from which control laws can be derived.