具有动作解析功能的情感交互机器人

仝佳媛^{1, 2}，谷清水^{1, 2}，左国玉^{1, 2}

1. 北京工业大学信息学部，北京市，100124;

2.计算智能与智能系统北京市重点实验室，北京市，100124

摘要：在现代社会中，人们对情感陪伴有越来越多的需求，机器人宠物为人们提供了一种很好的情感交互方式。本文设计了一种外观类似抱枕的具有动作解析功能的情感交互机器人，该机器人对抚摸动作进行情感分析，以识别人的各种情感抚摸及睡觉、拥抱等动作，机器人以不同的摇动尾巴和发声的方式对人的动作情感进行响应。本文首先从参数提取、动作情感的识别与分析、机器人情感反馈三个方面对机器人系统进行设计；其次，采用stm32控制器，实现机器人的人体动作感知、尾部舵机的动作表达、人机语音交互功能，构建虚拟机器人系统，并实现与真实机器人的尾部同步摇摆；最后采用BP神经网络的情感分析方法对人的动作进行情感分析并进行情感反馈。实验表明，所研制的实物系统及机器人情感分析方法在一定程度上证明了所设计机器人系统的有效性，达到了设计目标。

关键词：情感机器人；宠物机器人；情感分析；BP神经网络；系统设计

Emotionalinteractiverobotwithactionanalysisfunction

Tong Jiayuan^{1, 2}，Gu Qingshui^{1, 2}，Zuo Guoyu^{1, 2}

1. Beijing University of Technology, Faculty of Information Technology, Beijing City, 100124;

2. Beijing Key Lab of the Computational Intelligence and Intelligent System, Beijing City, 100124

Abstract：In modern society, people have more and more needs for emotional companionship. Robot pets provide a good way of emotional interaction. This paper designs a kind of emotional interaction robot that looks like a pillow and has action analysis function. The robot performs emotional analysis on stroking action, identifies all kinds of human emotional stroking, sleeping, hugging and other actions. The robot responses to human's action emotion in different ways, such as wagging tail and making sound. Firstly, this paper designs the robot system from three aspects：parameter extraction, motion emotion recognition and analysis, and robot emotional feedback. Secondly, the stm32 controller is used to realize the human motion perception, the motion expression of the tail steering gear and the human-computer voice interaction function of the robot. The virtual robot system is constructed and synchronized with the real robot's tail. Finally, BP neural network emotional analysis method is used to conduct emotional analysis of human action and emotional feedback. Experiments show that the physical system and the emotional analysis method of the robot prove the validity of the designed robot system to a certain extent, and achieve the design goal.

Keywords：Emotional robots;Pet robot;Emotional analysis;BP neural network;System design