The implementation of robots endowed with cognitive and social skills is getting a lot of attention in Social Robotics research. From the AI research standpoint, the symbiosis of humans with robotic agents in less controlled environments foregrounds a variety of research challenges, including understanding how to operate objects not previously known, communicating with humans with socially acceptable means of interaction, exhibiting predictable and common sense behavior, as well as learning the dynamics of open, less controlled domains.
SoCoLA agents will rely on their socio-cognitive skills, in order to formally describe potentially unknown domestic objects, such as the set of context-dependent actions that can be performed on them, the purpose of use of these objects in a home setting, and whether they are recommended for the person the agent is collaborating with.
Individual research objectives and contributions involve:
- The development of formal theories for enabling robots to learn both from observation and speech, building on the inherent correlation between these two modalities in social interactions.
- The creation of a novel argumentative means of interaction between robots and humans that will offer a breakthrough in existing dialectical interactions.
- The development of a framework that brings together actuation, common sense reasoning, dialectical interaction and knowledge extraction from the Web in a unified approach.
- The demonstration of the applicability of the overall approach structured around 3 types of interactions of increasing complexity, namely explanatory, exploratory and argumentative dialogues
