Robots in the classroom: Learning to be a Good Tutor

Abstract

To broaden the adoption and be more inclusive, robotic tutors need to tailor their
\nbehaviours to their audience. Traditional approaches, such as Bayesian Knowledge
\nTracing, try to adapt the content of lessons or the difficulty of tasks to the current
\nestimated knowledge of the student. However, these variations only happen in a limited
\ndomain, predefined in advance, and are not able to tackle unexpected variation in a
\nstudent's behaviours. We argue that robot adaptation needs to go beyond variations in
\npreprogrammed behaviours and that robots should in effect learn online how to become
\nbetter tutors. A study is currently being carried out to evaluate how human supervision
\ncan teach a robot to support child learning during an educational game using one
\nimplementation of this approach.