An Emotional Engine Model Inspired by Human-Dog Interaction Csan´ ad Szab´ o, Andr´ as R´ oka, M´ arta G´ acsi, ´ Ad´ am Mikl´ osi, P´ eter Baranyi, P´ eter Korondi Abstract— We use an engineering approach based on state space modeling to develop the emotional engine of an artificial agent. The model is inspired by the latest achievements of human-dog interaction research. Our goal is to build a model that can be easily integrated into existing applications in the field of robotics, computing and communication systems. The proposed model ensures this integration by using simple arithmetics, while being scalable and stable. I. INTRODUCTION In the past decades robotic research interest has broadened to include domestic applications, and huge efforts have been invested in building robots that support people in their daily life. Robots that are used in close collaboration with people face very different challenges and have to confront and solve new problems in comparison to industrial robots. Industrial robots work in well-known environments (factories), safety is guaranteed through special instructions and the robots receive their tasks via computers. In contrast domestic robots are required to work in environments that are often unfamiliar and very dynamic. Furthermore, these robots are expected to interact and cooperate with people directly. This requirement for human-robot interaction forces robot designers to deal with cognitive, emotive aspects in addition to the technical and technological issues in robotics. We refer to robots that interact and communicate with humans by following social behaviors attached to their roles as social robots. These social robots possess emotional models to ease their interaction with humans. The research of social robots integrates impacts from arts, engineering, medicine and other scientific fields. Social robots have been already applied in the fields of entertain- ment, healthcare, education, child- and eldercare. In our work we use an engineering approach inspired by a general model of basic emotions based on human-dog interactions to develop an abstract model of the emotional engine of an artificial agent. The emphasis of our work is on the theoretical side of the modeling. The challenge of the present project is to build a non-human emotional agent which is able to interact with humans in a natural way and is based on a simple engineering model utilizing simple mathematical methods. Our goal is to develop a model that can be integrated in numerous existing applications from social robots to personal data assistant equipment as well Csan´ ad Szab´ o, Andr´ as R´ oka, P´ eter Baranyi and P´ eter Korondi are with the Computer and Automation Research Institute, Hungarian Academy of Science, Budapest, Hungary {csanad.szabo,andras.roka,baranyi,korondi}@sztaki.hu M´ arta G´ acsi and ´ Ad´ am Mikl´ osi are with the Department of Ethol- ogy, E¨ otv¨ os University, Budapest, Hungary gm.art@t-online.hu, amiklosi62@gmail.com as ”simpler” domestic applications. The only requirement towards the robotic application is that it has to have ”instru- ments” to express its inner states which will be recognized as emotions on the human side. This paper is organized as follows. Section II gives an overview of emotional models applied in different social robots. Section III presents the methodology we used to set up our model. Section IV deals with the ethological background of our approach. Section V describes first the parts of the emotional system then the complete system which implements our emotional engine. Finally, Section VI concludes the paper and sketches possible future work. II. OVERVIEW OF USED EMOTIONAL MODELS Social robots using emotional models have a general structure of emotion handling components (cf. Fig. 1). The robots are equipped with input devices (visual, audio, me- chanical sensors) to recognize events in the environment. Identified events carrying emotional information are passed to the emotional engine. The emotional engine interprets the information and computes the impact of the event in the emotional model. Finally, the emotional engine interprets the emotions identified in the model and makes a transformation for the output peripheries of the robot in order to express the emotions. Fig. 1. General structure of emotion handling components in the robots In order to identify the relationship between our emotional model and existing emotional models used in social robotics, we review six robots with regard to their abilities, with special focus on the used emotional models. Figure 2 shows an image of each of the investigated robots. Fig. 2. Investigated robots: Ifbot, Probo, Aibo, iCat, Kismet, Emotirob (from left to right) We set up several evaluation criteria to investigate emotion handling of the robots from our point of view. Firstly, we investigated whether the emotional model distinguishes between the duration of emotional states (short-, medium-, long-term). Secondly, we searched for probabilistic methods used in the context of the emotional model. Finally, we