Understanding emotional values of bionic features for educational service robots: A cross-age examination using multi-modal data

With the rapid development of artificial intelligence and robot technology, social robotics are increasingly integrated into people's daily life. Among them, an emerging application domain is robots for assisting teaching and learning. The current developments of educational service robot (ESR) pay much attention on the bionic design of robots to fulfill users’ affective preferences for improved using experiences. Also, students at different age groups exhibit varied opinions towards the same robot design, which influences the acceptance of ESR. In this paper, we focus on measuring how different bionic features of ESR influence users’ emotion and analyzing the resulted differences between secondary school students and college students. Specifically, multi-modal data are collected with 28 participants (14 secondary students vs. 14 college students) using electroencephalography (EEG) and eye-tracking (ET) technology, to examine the influence of different ESR bionic design (animal-shaped, humanoid and abstract-shaped) on affective preference of users. Based on the obtained data, we try to compute a psychophysiological model for customized design decisions, through studying the relationship between psychological value and physiological features of the participants. In order to improve the accuracy of the psychophysiological model, we investigate an improved support vector machine (SVM) based on a multi-kernel learning to map physiological features of different modalities with psychological values of the participants. Our statistical analyses of study results show that there are significant differences between secondary students and college students in their interest for different ESR bionic features. Additionally, the proposed psychophysiological model presents higher prediction accuracy, better generalizability and robustness than typical machine learning models.