Human-Robot Interface (HRI)

KontaktOrhan-Can Görür, Dr. Axel Heßler

Human Robot Interaction (HRI) has become one of the most popular research field with the recent developments on the robots. The robots are now robust and skilled enough to be placed in the environments where their interactions and collaboration with humans are inevitable (e.g. assistant/companion robots in home, industry, education, caring etc.). Through their acceptance in the society and thrustworthy collaborations with humans, the robots need to exert human-like reasoning and behaviours. Having the purpose of building such robots, our approach to HRI is through social intelligence in robotics. The vision is to develop human-like cognitive abilities in robots, i.e. their understanding of the human behaviours, adapting to individuals (personalized robots) and responding better towards meeting individual’s needs.

In DAI-Labor, the researches on HRI are in the context of application projects, Bachelor, Masters and PhD theses. Our research in particular focuses on developing a computational cognitive model for the robots, which we divided into two processing layers. In the bottom-layer, we focus on cognition abilities such as: Human presence detection; face (offline and online), speech and gesture recognition; pose estimation and tracking; attention detection and environment detection. Whereas in the top-layer, these low-level skills are used in further human-like reasoning to better understand emotional and motivational states of the person. In particular, the top-layer focuses on on-line machine learning methods to learn long-term emotional cues of the person (e.g. moods, desires, preferences etc.) by adapting to individual’s recognised behaviours and responses. This level of reasoning in social robots will autonomously answer the question: “How to actually respond to the expressed behaviours of a specific person?” leading to decisional autonomy for the robots in HRI.