Research

AI for social interaction

In the Social AI Lab we research artificial intelligence for social interaction between man and machine, using smart displays, social robots, and wearables. We develop a Social Interaction Cloud integrating speech and vision technology for analyzing social cues from and shape the dialogue with human actors. The lab facilitates the evaluation of our social interaction designs for, e.g., education, health care, and retail and interdisciplinary cooperation with the social sciences.

Contact: Koen Hindriks

Health Behaviour Informatics

This research takes a systems approach to design and evaluate computational health behaviour systems. Health behaviour informatics aims to address the full cycle of automatically measuring, analysing, and acting upon the physiological and psychological state of individuals. Matters are complicated by the fact that most observed behavior will emerge from unknown or poorly understood sources, requiring adaptive AI systems.

Contact: Aart van Halteren

Human-robot teaching and learning

This research direction is concerned in developing computational methods for flexibly transferring information and skills between humans and robots through social interaction. This transfer can occur in the context of explainable robot behavior (robot-to-human transfer of information), assistive settings (robot helping human or human helping robot to get to a goal), or through teaching interactions (interactive robot teachers or learners).

Contact: Kim Baraka

AI for social simulation

Within my research I aim to enrich computational agent based models with knowledge from the social sciences in order to make these models more realistic. The concept of agent-based modelling forms the key method. Hereby, I distinguish three closely connected research lines, which are centered around three sub-areas of agent technology, namely agent-based social simulation, agent-based predictive modelling, and intelligent virtual agents.

Contact: Charlotte Gerritsen

AI for analyzing and modelling conversation

Conversation is an ideal interface to accomplish tasks and achieve common understanding, whereas the richness of language and context poses challenges to properly interpret a user’s utterance and decide on how to proceed in the conversation. We take up this challenge in case studies on interpreting messages in user fora, as well as in developing a cooking assistant and patient interview robot, combining Natural Language Processing, conversation analysis, cognitive agent technology and knowledge reasoning.

Contact: Florian Kunneman