Emotion Estimation Using Body Expression Types Based on LMA and Sensitivity Analysis

摘要

The authors have been working on methods of estimating emotions from spontaneous body movements during a collaborative fabrication task. In this study, the authors propose a novel method of grouping emotional expressions on the basis of body movements to create type-specific emotion estimation models. We assume that there are typical expression types that are idiosyncratic characteristics for each person. We propose a method of grouping people by expression types using body movements only, based on Laban Movement Analysis (LMA) and the concept of sensitivity analysis. We perform an experiment to measure body movements and emotions during a fabrication task. Finally, we evaluate the accuracy of estimation and type-specific estimation models. The accuracy was, on average, 67.92%, which was 18% higher than that of person-independent models and superior to existing studies.