<i>KANSEI</i> GENERATOR USING HMM FOR VIRTUAL <i>KANSEI</i> IN CARETAKER SUPPORT ROBOT

摘要

This study proposes a new method of constructing the KANSEI (emotions, feelings and affective behaviors) engine using Hidden Markov Model (HMM). The KANSEI engine aims to generate human-like emotion transition. HMM is suitable for nondeterministic time-series data such as emotion transition of human. This paper describes three types of proposed engines using the HMM technique and compares for their performances. One of the engines was also compared with the conventional KANSEI engine using Petri-Net with Genetic Algorithm. The experiment showed that the proposed engine performed better than the conventional KANSEI engine.