Cognitive Interaction Technology

While there are amazing solutions for highly specialized action generation in narrow domains, the synthesis of similar capabilities for unprepared, everyday environments is a major unsolved challenge. Here, major research foci are to consider the control of highly redundant, anthropomorphic actuator systems, such as full-body movements and multifingered hands. Biology, psychology, sports science and robotics groups interact closely to combine algorithmic approaches with insights gained from the analysis of human and animal motion and to come closer to a coherent picture about the internal representation of our movement abilities. A connected research objective is to synthesize robot motions that match better to the expectations of humans in order to achieve smooth cooperation.

Research in this area is directed at mechanisms that enable a system to understand and actively focus on what is important, to ignore irrelevant detail, and to share attention with humans, including the important ability of “being social” by providing intuitive, “emotional” feedback (Fig. 1) in order to make robots gain better acceptance and to be understood in a more natural and direct manner. Major goals are to create novel perspectives on attention mechanisms that integrate computational, psychological, and social dimensions, and to create new experimental methodologies for exploring and testing hypotheses about attention mechanisms. To this end, the participating research groups explore the implementation of attention mechanisms at different levels and on a wide spectrum of technical platforms ranging from display-bound attentive user interfaces over augmented reality systems to virtual agents or embodied anthropomorphic robot systems.

This is a type of communication where the use of language is heavily drawing on the simultaneous presence of shared information in non-linguistic modalities, most notably vision. To elucidate the factors that give situated communication its superb economy and flexibility and to create the necessary integrated perspective for a deeper understanding, linguists and computer scientists are closely collaborating, combining corpora-based research, modeling, theory building, and the exploration of technical systems. From a technical perspective, an important line of research is the exploration of innovative approaches for enhancing natural and artificial styles of communication by combining visualization, sonification, haptic, and augmented reality devices in novel ways. Fig. 2 depicts a communication situation where configurations of natural objects are linked to data displays made available through AR headsets. Such settings offer novel ways to study how communication is affected by conflicts in situational contexts which may be injected in a highly controlled way through the AR devices.

Memory connects perception with experience and allows cognitive processing to generate intelligent action. A major goal is to develop memory systems that can approximate some of the key features of human memory, such as flexible association, scalability and learning at different levels. This raises questions such as: what are the required representations and how do they have to interact with each other? How can we integrate new pieces of knowledge, and what might be the use of forgetting in a technical system?

To address these issues, neuropsychologists and biologists groups working on the brain and its architecture share their insights with sports scientists and robotics researchers. They combine neuropsychological research on human memory with methods for analyzing memory structures from interviews and behavioral studies. This work is complemented by computational approaches exploring biologically inspired memory architectures and combining techniques from the field of datamining with learning concepts ranging from statistical machine learning, neural networks, the extraction of ontologies with symbolic methods up to biologically inspired approaches of imitation learning. Another important aspect are developmental studies, analyzing the development of infant cognition as well as communicative principles of parent-infant interaction during the learning-by-demonstration process in order to gain novel insights for technical systems, and the interconnection of different modalities to maintain the linkage between conceptual and perceptual levels to enable coherent behavior.

Of the current groups, twelve are belonging to the founding principal investigators Their groups plus two newly created computer science groups (Computer Graphics/Mario Botsch; and Text Technology/Alexander Mehler) bring in a base personnel of about 100 project scientists. Together with the new groups described below, they are supported by an administrative office with a general manager and adjunct managers responsible for the areas public understanding of science, gender issues and financial administration. The scientific affairs of CITEC are directed by a scientific board meeting regularly to coordinate the cluster.

A substantial part of CITEC funding has been invested in the creation of new groups. Three new Computer Science (TEC) groups on along with five newly established young investigator tenure-track groups with a last young investigator group on Neuromorphic Engineering in the process of becoming filled.

Additionally, a considerable number of reappointments have been or are currently closely coordinated with the research objectives of CITEC and are partially supported with CITEC-funding: one biology group (V. Dürr—Biological Cybernetics), two linguistics groups (J.P.de Ruiter—Psycholinguistics, P. Wagner—Phonology) and one psychology group (W. Schneider—Cognitive Psychology). Three further re-appointments (PSY, successor of G. Draisbach, BIO, successor of Y. Winter, and Philosophy, successor of A. Beckermann) are currently negotiated.