skip to main contentskip to main menuskip to footer Universität Bielefeld Play Search
skip breadcrumb navigation to main content
  • Medizinische Fakul­tät OWL

    Grafik vom Innovationszentrum Campus Bielefeld
    © Universität Bielefeld

Modelling the multimodal dialogue in co-constructive task learning

Switch to main content of the section

Prof. Dr.-Ing. Britta Wrede

Professur für Medizinische Assistenzsysteme

Manuel Scheibl

Wissenschaftlicher Mitarbeiter

Dr.-Ing. Birte Richter

Wissenschaftliche Mitarbeiterin
AG Medizinische Assistenzsysteme

Project duration

August 2023 - today

Logo co-constructive task learning
© Bielefeld University

Research on communication between parents and children has shown that parents in a teaching scenario adapt the strategy of the conversation to the level of understanding of the learner. A key factor in this is the adjustments made in response to the child's attempt at imitation. Emphasising relevant aspects of a movement or task, as well as additional verbal explanations of details, are possible such adaptations. Another important aspect is the association with common past experiences that can be recalled by the child.

A purposeful learning scenario also requires that the child is able to recognise his or her gaps in knowledge and ask specifically for missing information or contexts that are missing for correct imitation. A dynamic interplay of learning is thus created between the teacher and the pupils, shaped equally by both parties [Centre for teaching and learning].

Our goal is to develop a framework that allows such a frame of interactions (Pragmatic Frame) between a human and a robot in a learning scenario.
It is relevant for the intuitiveness of the learning scenario that all natural means of communication, such as gesticulative and verbal communication, can be used.
 

How the different input modalities can be stored in a uniform format and made semantically retrievable are the crucial questions to be answered on the way to the Pragmatic Frame. The appropriate link to given detailed information is just as important here as the mapping of simultaneous information influences such as verbal and gestural input. Finally, the intuitiveness of the developed framework is also relevant and will be investigated.
 

This project is part of the Co-Constructive Task Learning project.

Click here for Manuel Scheibl's Gitlab repository.

Systems

back to top