Design of an Architecture Based on Emotions for an Autonomous Robot

With the emergence of personal robots recently the need to create an easier form of human-robot interaction has become apparent. Emotions will play an important role in this interaction. In the proposed architecture emotions are generated from the evaluation of the wellbeing of the robot. Moreover, behavior selection mainly depends on a specific drive and emotions have a supervisory role. Proposed Architecture The goal of our research is to construct autonomous personal robots. These robots are being built on mobile platforms (iRobot Magellan Pro), on which various sensors (laser, sonar and camera) and devices are installed. These sensors and devices will make possible the interaction of the robot with its environment and users. Fig. 1. AD Architecture with an emotional control system. Figure 1 illustrates how an emotional control system (ECS) has been added to the AD architecture (Barber and Salichs 2001). Inside the ECS we can observe three different modules: Drives, Activity Selection and Emotional Supervisory System (ESS). The Drives module is the one that controls the basic drives of the robot. The Activity Selection module on the other hand, determines goals and action tendencies of the robot. Finally, the ESS module generates the emotional state of the robot. Drives Several authors, such as Frijda (Frijda 1987), propose that an independent system should not have to wait for someone to maintain, succour, and help it. Therefore, an autonomous agent should be capable of determining its goals, and it must be capable of selecting the most suitable behaviour in order to reach its goals. Similarly to the Velasquez’s proposal (Velasquez 1998), the robot’s autonomy relies on a motivational model based on drives and emotions. Within the model developed by Velasquez, drives are the robot’s internal needs. These drives are related to the feedback of internal variables such as its battery and temperature. By contrast, within our control architecture the drives are defined in a different way. They are related to the feedback of internal variables (e.g. battery level) as well as external variables (e.g. social interaction). In this architecture the following drives are initially used: Energy: With the battery level as its input. Entertainment: Depending on the activity and humanrobot interaction. Affect: Depending on the level of affection received through the human-robot interaction. Health: Related to physical damage. Agenda: This drive reflects the motivation of the robot to carry on with activities which have been programmed previously. As it is shown, social motivations are included as robot’s needs. The Entertainment and Affect drives are similar to the Stimulation and Social drives defined by Breazeal (Breazeal 2002). Emotions Most of authors working with emotion-based control architectures, such as Velasquez (Velasquez 1998), Gadanho (Gadanho 2002), Breazeal (Breazeal 2002), etc., give a basic role to emotions i.e. behaviour selection is conditioned by emotions. Nevertheless, according to several authors such as Ortony (Ortony, Clore and Collins DELIBERATIVE LEVEL AUTOMATIC LEVEL Long term Memory