An Integrated System Mimicking Brain Information Processing with Learning Functions for Self and Others, and Its Application

本研究开发了以下五种模型：（1）预测其他人的状态和动作的状态-动作预测模型，（2）估计其他人的动作策略的策略估计模型，（3）估计其他人的意图的意图估计模型，（4）关注多个敏感刺激中的特定刺激的注意力产生模型，（5）模仿人类情感的情感生成模型。在此基础上，基于提出的五种多智能体协作行为模型，开发了一个具有学习功能的模拟大脑信息处理的集成系统。通过计算机仿真和机器人实验，对所提出的系统进行了评估。

This study developed the following five models: (1) a state-action prediction model which predicts other's state and action, (2) a policy estimation model which estimates other's action policy, (3) an intention estimation model which estimates other's intention, (4) an attention generation model which focuses on particular stimulus out of plural sensitive stimuli, (5) an emotion generation model which imitates human emotion. After that, an integrated system mimicking brain information processing with learning functions was developed based on the five proposed models for emerging cooperative behavior in multi-agent systems. Through both computer simulation and robot experiment, the proposed system was evaluated.