Interaction Design Using a Child Behavior-Geometry Database

摘要

Unintentional injuries, that is, injuries due to accidents, account for a large share of the cause of death in children. Some accidents can be prevented by designing products that are based on an understanding of the interaction between children and the object. Improving products to prevent injuries requires a system that helps product designers predict the behavior that the object induces in children. In this paper, we developed a behavior-geometry database that consists of 1) data on children's behavior with common objects, 2) for various objects, data from 3D shape models for which the feature vectors are calculated by a 3D discrete Fourier transform, and 3) two kinds of models for using a 3D shape-feature vector to predict the induced behavior, the barycentric behavior model and the multiple linear regression model. We also developed the following behavior-symmetry-search functions that use the database: a) a shape-similarity search, b) an induced-behavior search, which is a function for predicting the behaviors induced by an object's 3D shape, and c) a behavior-symmetry search, which is a function for finding objects that induce behaviors similar to those induced by the shape of a target object. The third function is useful for finding shapes that are similar in terms of inducing child behavior. In this study, we evaluated the effectiveness of the implemented system using data from 275 accidents and 3D shape data from 45 objects.