Towards Adaptation and Personalization in Task Based on Human-Robot Interaction

Abstract

Social robots increase peoples curiosity and motivation in social and intellectual activities because people feel challenged to interact with machines and find out how smart they are. Such systems are being built to work in entertainment environments, homes, hotels and assistive tasks. However, a decrease in the user motivation and attention span is noticed after the robot loses its novelty. This paper describes a proposal of a dynamic user adaptation system applied to a humanoid robot as a framework to create activities and autonomously interact with the users, prolonging the interaction time. Machine Learning methods are used to detect and classify the verbal answers of the users and body signals to adapt the robot behavior. The system stores information of the users to personalize some dialogues during the activities, trying to simulate rapport building. A storytelling activity was programmed to test the proposed system. Experiments performed showed that the participants were able to perceive the robot's behavior adaptation and also the personalization impact along the interaction session.