User modelling for personalised and co-adaptive human-robot interaction

Simulating the tremendous social adaptation abilities that characterise human interactions requires the establishment of bidirectional processes in which humans and robots synchronise and adapt to each other in real-time by means of a mutual exchange of verbal and non-verbal behaviours in order to achieve mutual co-adaptation. The ability to respond in a contingent manner to users’ needs, preferences, interests, intentions and emotions is, in fact, of paramount importance to achieve long-term robot autonomy, that is, to establish and maintain autonomous interactions with human users over extended periods of time. To endow robots with co-adaptation abilities, a typical approach leverages statistical learning to incrementally adapt robot’s behaviours and strategies to a specific user’s situation, for example, a user’s emotion, personality or progress with a task. This affect-based, personalised co-adaptation aims to close the human-machine loop while enabling robot learning from human users in more natural, intuitive ways.

Selected Publications

Jones, A., & Castellano, G. (2018). Adaptive Robotic Tutors that Support Self-Regulated Learning: A Longer-Term Investigation with Primary School Children. International Journal of Social Robotics, 10(3), 357-370. [PDF]

Castellano, G., Leite, I., & Paiva, A. (2017). Detecting perceived quality of interaction with a robot using contextual features. Autonomous Robots, 41(5), 1245–1261. [PDF]

Castellano, G., Leite, I., Pereira, A., Martinho, C., Paiva, A., & McOwan, P. W. (2014). Context-Sensitive Affect Recognition for a Robotic Game Companion. ACM Transactions on Interactive Intelligent Systems, 4(2). [PDF]