SMILE AND TALK - ANA, THE ARCHITECTURE OF A ROBOT THAT VERBALLY AND NONVERBALLY UNDERSTANDS YOU

Abstract

Robots capable of recognizing and analyzing verbal and nonverbal communication can play important role in many domains, especially where the verbal communication does not provide enough information to build the best interaction experience. The ability to remember a person or to analyze his/her facial expressions can be important tools to improve the human-robot interaction, especially when the robot might have to deal with different people for long time. In this work it is presented the interaction architecture of ANA, the robotic receptionist created to interact with Brazilians speakers. The architecture pro-vides support to verbal communication, such as voice and text inputs, and nonverbal information, such as facial and basic emotions recognition. To increase the robustness of interactions, the knowledge of the robot was modeled over ontologies using local and remote bases. To validate the model, ANA interacted with workers, provided information and captured their emotions and classified them. These sentiments were used to calculate the emotion average rate for each hour of the day, providing data about the workers' behaviors.