Proposal of Human-like Spatiotemporal Language Understanding Based on Mental Image Model for Language-centered Human-Robot Interaction

Abstract

Mental-image based spatiotemporal (4D) language understanding in human was considered from the viewpoint of cognitive science and simulated based on the mental image model proposed in mental image directed semantic theory (MIDST). The application system was evaluated based on a psychological experiment involving human subjects and found a good success in answering questions concerning the scenes expressed in 4D language.