TU München: Cluster of Excellence Cognition for Technical Systems (CoTeSys)
They differ from other technical systems in that they perform cognitive control and have cognitive capabilities. Cognitive control orchestrates reflexive and habitual behavior in accord with longterm intentions. Cognitive capabilities such as perception, reasoning, learning, and planning turn technical systems into ones that "know what they are doing". More specifically, a cognitive technical system becomes a technical system "that can reason using substantial amounts of appropriately represented knowledge, learn from its experience so that it performs better tomorrow than it did today, explain itself and be told what to do, be aware of its own capabilities and reflect on its own behavior, and respond robustly to surprise". Technical systems that are cognitive in this sense will be much easier to interact and cooperate with, be robust, flexible, and efficient.
In order to design, realize, and analyze the information processing mechanisms needed for cognitive technical systems, the CoTeSys cluster combines the research competence in neuroscience, natural sciences, mathematics, engineering, computer science, and humanities from the Technische Universität München (TUM), the Ludwig-Maximilians-Universität (LMU), the Universität der Bundeswehr (UBM), the Deutsche Luft- und Raumfahrtzentrum (DLR) and the Max-Planck-Institut for Neurobiology (MPI). Fundamental research in neurobiology, neuroscience, and cognitive science form the basis of novel engineering and computing approaches to cognition. Three demonstrative scenarios are pursued: cognitive aerial and terrestrial vehicles, cognitive humanoid robots, and cognitive factories. CoTeSys organizes and structures the intended interplay of disciplines, institutes, research foci, and demonstration testbeds in the following way.
Foundational research is conducted in the research area "Neurobiological and Neurocognitive Foundations" by studies of humans and animals at the behavioral and brain level. The goal is to produce computational models of perception, motor control, and learning. These models and the respective insights of how natural cognitive systems achieve robustness, adaptability, and performance are then used as inspiration and resources to realize information processing mechanisms that will accomplish the different cognitive capabilities.
The research areas "Multisensory Perception for Action", "Knowledge and Learning", "Action Planning and Control/Joint Action", and "Human Interaction" investigate individual cognitive capabilities. The scope of the areas are defined with respect to the kind of cognitive capability that they realize. There are two key driving forces for the research: Firstly, the insights and models developed in the foundational research area in natural and artificial cognitive capabilities. Secondly, the challenges that are generated by the demonstration scenario research raising applied and basic research questions.
The demonstration platforms and scenarios challenge fundamental research, provide an essential platform for interdisciplinary cooperation, and are the stage for CoTeSys engineering research.