We propose to research how virtual characters can be used to enhance the emotional experience people have in virtual group interaction. In particular, we experimentally investigate what aspects and parameters impact how a virtual character induces emotions during an online group activity.

For this, we plan to (1) record (audio and video) a data set of people’s reaction and responses in a virtual meeting, (2) label the data set, and (3) do an initial analysis of the dataset to inform the design of emotionally intelligent virtual characters. We would try to automate data collection as much as possible to make it easy to add more data over time.

This will contribute to a longer-term effort in model and theory formation.

The main Contribution is to WP3. This is set in Task 3.2: Human-AI Interaction/collaboration paradigms and aims at better understanding shared emotion online setting.

Output

dataset of people’s reactions onces it is their turn (e.g. 10 second video, audio) and the information about the type of turnover (1-5) that took place before

a manuscript describing the data set and initial insights into creating intelligent agents that induce positive emotions

Project Partners:

  • Ludwig-Maximilians-Universität München, Albrecht Schmidt
  • LMU, Albrecht Schmidt
  • DFKI, Paul Lukowicz and Patrick Gebhard

Primary Contact: Albrecht Schmidt, Ludwig-Maximilians-Universität München