Enhancing Museum Visitor Engagement: Personalized Learning with Adaptive Robot Tutor

The present study describes the development and evaluation of an adaptive question selection system for a robot deployed in a museum that aims to test visitors’ familiarity with various robot types. The robot initially asked visitors questions randomly and recorded their responses; the analysis of more than 1,000 interactions showed that different robot types were recognized at different success rates. A Q-learning-based reinforcement learning model was then trained, enabling the robot to select questions adaptively depending on perceived difficulty. Initial findings point to the possibility of adaptive question selection to improve game outcomes and visitor engagement, with an almost 43% increase in the number of desired game outcomes.