Design and implementation of sociable robot for multimodal affective communication

Abstract

Emotion plays a vital role in human interaction. To interact sociably with a human, a robot has to recognize and express emotions like a human. It also has to speak and determine its autonomous behavior while considering the emotional status. This paper presents the Affective Communication Robot that is designed to recognize human emotion and express its emotion through multimodal emotion channels like a human, and behave appropriately in response to human emotions. Contrary to previous sociable robots that behave depending on its own emotion, the robot behaves in consideration of user`s emotional status. We improved the humanoid robot, AMI to be an affective communication robot in this paper. To do so, we designed and implemented the affective communication framework, which comprised of the five sub-systems: a perception, motivation, memory, behavior and expression system. In the perception system, we developed a bimodal emotion recognizer for emotion recognition. The other systems were tightly connected to determine an appropriate behavior to user’s emotional status with its own drive, emotion, and memory. As a result, AMI chose an appropriate conversation topic and behaved appropriately in response to human emotions. It could ask what the human is angry about and then make a joke to console him or give pleasure to him. Additionally, we developed a new robot, ACR for better interaction. Therefore, these robots were able to help human mentally and emotionally as a robot therapy function. The human partner perceived the robot to be more human-like and friendly, thus enhancing the interaction between the robot and human.