Event Representation and Semantics Processing System for F-2 Companion Robot

Abstract

F-2 companion robot is designed to implement and test various cognitive functions linked with text comprehension, as well as verbal and nonverbal communication strategies. F-2 has a syntactic parser and text comprehension engine, based on productions, where each incoming sentence meaning, or a computer vision event is associated with the most relevant scripts. The script engine is designed to simulate communicative reactions, emotional dynamics, and rational inferences. Scripts are activated, depending on the state of the emotion model, and provide output behavioral packages in Behavior markup language (BML), executed by the robot. We demonstrate simultaneous responses of the robot to the incoming phrases, human gazes, and events in the Tangram puzzle game, where the robot guides the player and emotionally reacts to the game events.