Hybrid emotion generation architecture with computational models based on psychological theory for human-robot interaction

Abstract

According to the expansion of the robotic application, the necessary role of humans in the interactions with robots has also diverged. Humans usually play the role as bystanders in service scenarios and social interaction capabilities are required in the interaction. Emotions also serve as a type of media to communicate each agent`s intention indirectly, modulate interaction, and construct a social relationship. Therefore, solving the problem of how to make an emotional interaction between a human and a robot is necessary for the realization of natural and social interaction between humans and robots. For this purpose, the robot should be able to generate its own emotions. The mechanism of emotion can be seen as an information processing system. The goal of information processing in the emotion generation mechanism is to make a robot be able to show suitable emotional reactions in relation to current situations while the robot performs its task and responds to any stimulus that the robot can sense. Emotions regarding current situations and emotions for sensations require processes that have different characteristics. The generation of emotions regarding a task-related event requires the analysis of relationship between the task and current situation. Because this process is a knowledge-based process, it can be considered as a deliberative process. Additionally, it is possible to apply the cognitive appraisal theory of emotion in this process to generate emotion. On the other hand, emotional reactions to sensations require a pattern-driven process. The perception of each sensation`s representative characteristics triggers a pre-defined emotional reaction, and the process can be defined as a collection of simple rules. Therefore, emotional reactions to sensations can be designed as a reactive process. Two different processes can be integrated in hybrid architecture. The hybrid architecture composed of a reactive and a deliberative layer is proposed to achieve the...