Research group
Social Cognitive Systems
Head
Prof. Dr.-Ing. Stefan Kopp
human abilities for
social interaction
from
multimodal communication
feedback behavior
adaptation mechanisms
imitation+
learning
...
agents, avatars, robots with similar skills
to
Methodology
1. Interaction studies and corpus-based analyses
2. Building computational models of cognitive and behavioral mechanisms
3. Evaluation and application in HCI scenarios Audience design of communicative behavior
Research question: Do humans gesture differently towards a computer vs. robot/character vs. human?
Task: Conduct a comparison study
• Manipulation: Type of addressee (computer vs. robot/character vs. human)
• Measures: Gesture rate, physical gesture form (e.g. size), gesture semantics etc.
Contact: Stefan Kopp <[email protected]>
How to make sense of system errors?
Research question: How do users deal with system-sided errors? How do errors affect future human-computer/agent interactions?
Task: Conduct a study with different kinds of interaction errors
(in comparison with human-human interaction data)
• Manipulation: Error patterns, degree of system anthropomorphization
• Measures: Human users’ adaptation and correction strategies
Contact: Stefan Kopp <[email protected]>
Computational Pragmatics
Context: Project develops a model that allows an agent to recognize the user’s communicative intention from her behavior and respond appropriately
Overall task: Recognition and interpretation of social signals for in multimodal HCI
Specific questions:
• Analyzing gaze behavior in natural human-human interaction data
• Building models of user behavior/signals
Contact: Sebastian Loth <[email protected]>
Modeling gaze behavior in social interaction
Research context: How can we estimate others’ mental states? (Theory of Mind)
Task: Development and evaluation of a cognitive model of social gaze behavior
Specific questions:
• Intentionality: Is a potential interlocutor addressing me?
• Social gaze: What kind of gaze helps to solve this problem?
Contact: Sebastian Kahl <[email protected]>
Computer Animation of Believable Virtual Character Behavior
Research question: How to animate believable, expressive virtual characters?
Goal: Combination of different animation techniques into fluent and coherent agent behavior
Focus: Integrate Motion Capture with procedural animation, blending,
manage conflicts between animations/behavior, extension of/integration into
ASAP framework
Contact: Stefan Kopp <[email protected]>
Interactive visualization of Bayesian networks
Context: Making Bayesian inferencing available for different projects (PRIMO toolkit)
Tasks:
• Extend or redo the existing web-based visualization prototype
• Improve or extend the existing PRIMO backend
Useful skills:
• Basic knowledge of Bayesian Networks and probabilistic inference
• Experience with web-based visualization (e.g. d3.js)
Contact: Jan Pöppel <[email protected]>
Extending a Probabilistic Reasoning and DecisionMaking Framework
Context: Making Bayesian inferencing available for different projects (PRIMO toolkit)
Tasks:
• Add support for various kinds of continuous distributions
• Extend upon existing learn or inference algorithms
• Add support for different kinds of networks (e.g. Conditional Random Fields etc)
Useful skills:
• Decent knowledge of Bayesian Networks and probabilistic inference
• Proficient knowledge in probability theory
Contact: Jan Pöppel <[email protected]>
Further information:
https://scs.techfak.uni-bielefeld.de/
theses/open-theses