Behavior selection and motion modulation in emotionally grounded architecture for QRIO SDR-4X II

Abstract

This paper focuses OD the design and evaluation of behavior module selection and motion modulation based on emotion in the emotionally grounded (EGO) architecture that is applied in autonomous robot QRIO SDR-4 X II. In the EGO architecture, a behavior module is selected based on homeostasis in order for the robot to regulate its internal state within a certain range. Each behavior module has a value called activation level (AL) that is composed of motivation value (Mot) and releasing value (Rel). Mot determines the degree that the instinct drives the behavior module. It is derived from internal state. Rel is the degree that reflects how much external stimuli would satisfy an internal state as a result of the behavior. It is derived from the internal state and external stimuli. Each behavior module competes in AL to select a behavior module. Emotion is a key for motion modulation. It is derived from the internal state, its change, and the expected change of internal state associated with external stimuli in long-term memory. Through the implementation and experiments on QRIO SDR-4X II, we confirm the behavior selection and motion modulation processes in the EGO architecture.