Which mutual-modelling mechanisms to optimize child-robot interaction
Utku Norman is a doctoral assistant at CHILI Lab, and a PhD student in Computer Science (EDIC), Swiss Federal Institute of Technology in Lausanne (EPFL), Switzerland. He is driven by a desire to help build a better future, and understand the world along the way: His chosen course for how is advancing machine intelligence to develop systems that try to understand us. Utku obtained his MS in Computer Engineering from Ciceklab at Bilkent University, Turkey, where he developed a graph-theoretical model of neural development that incorporates information in time and space in order to discover risk genes for autism. He holds a BS in Electrical and Electronics Engineering, and a BA in Philosophy (double major) from Middle East Technical University, Turkey. Besides, he gladfully spent a semester in Technical University of Denmark (DTU) as an Erasmus exchange student. Utku enjoys learning languages, travelling and reading, as well as petting his dog: Çörek (Patty).
Ecole Polytechnique Fédérale de Lausanne (EPFL)
Pierre Dillenbourg (EPFL), in association with Ginevra Castellano (UU) and Chloé Clavel (IMT)
IMT; UU
In the context of collaborative learning with a robot, the ability to establish a mental model of the other is crucial in order to interpret and respond in an appropriate manner. As humans, this skill of mutual modeling is performed during most interactions by attributing goals and beliefs to others. The aim of this ESR project will be to investigate ways to enable a robot with mutual modelling in a collaborative learning task with a child. The model will investigate the impact of the strategies proposed for mutual modelling on engagement, and motivation of the child in the learning task as well as potential learning gain. After reviewing the literature in mutual modeling, engagement and motivation, the ESR will investigate methods for collaborative learning in human-human interaction and co-learning with robots. The aim will be to propose strategies to motivate and engage the child in co-learning while maximizing the learning gain. The proposed approaches will be evaluated via real experiments with children in learning contexts.
Completed draft of PhD dissertation, software and algorithms, peer-reviewed publications, international journal and conference publications