Prototype and Evaluation of A Guiding Robot with Continuous Training at Exhibition

Abstract

This study was made to increase user who uses a guiding robot. A robot should not guide at random, and it is necessary to select people with interaction willingness and do good interaction. Because there are many unclarified elements in interaction, a robot learns good interaction from previous experience. Therefore, we apply The Situation Scenting Model that can estimate human willingness and train online its model parameters to learn interaction to a guiding system. As a result, the ratio of the number of interactions to participants of the exhibition increases by 12%. Finally, we confirmed the proposed method valid for increasing guiding robot users.