Behavioral Accommodation towards a Dance Robot Tutor Raquel Ros, Alexandre Coninx, Yiannis Demiris Imperial College London, UK {raquel.ros, a.coninx, y.demiris}@imperial.ac.uk Georgios Patsis, Valentin Enescu, Hichem Sahli Vrije Universiteit Brussel, Belgium {gpatsis, venescu, hichem.sahli}@etro.vub.ac.be ABSTRACT We report first results on children adaptive behavior towards a dance tutoring robot. We can observe that children behav- ior rapidly evolves through few sessions in order to accom- modate with the robotic tutor rhythm and instructions. Categories and Subject Descriptors H.1.2 [Models and Principles]: User/Machine Systems; I.2.9 [Robotics]: Commercial robots and applications Keywords interaction adaptation; child-robot interaction; dance 1. INTRODUCTION Social interaction can be defined as a dynamic sequence of social actions between individuals (or groups) where in- teractors modify their actions and reactions based on the each other. Some examples of interaction adaptation that take place when people interact are matching other’s behav- iors, synchronizing the timing of behavior, or even behave in dissimilar ways [2]. Within the HRI community, most effort has been put on researching strategies to adapt the robot behavior to the user. In other words, considering the adaptation of a single interactor, instead of both. Though indeed it is a major goal to reach, we believe that taking into account the user capa- bilities and strategies to accommodate to the robot should be also considered and studied more deeply. Most research on adaptive interaction has been focused in conversational interactions, such as the Communication Ac- commodation Theory (CAT) developed by Howard Giles, where it is argued that people adjust speech, vocal pat- terns and gestures to accommodate to others. Following these ideas, linguistic adaptation of the user towards au- tomatic systems, both in computer-interaction and robot- interaction, have been observed in the past [4, 1, 3]. How- ever, little attention has been payed to behavioral (action) Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org. Copyright 20XX ACM X-XXXXX-XX-X/XX/XX ...$15.00. request (robot) response (child) interpretation ask ok performs correctly fail fails performing either creating its own motion or copying the robot’s motion observe observes and waits for robot’s ac- tion distracted does not follow robot’s instruc- tions show copy perfoms correctly copying the robot’s motion fail fails performing either creating its own motion or copying the robot’s motion observe observes robot’s motion distracted does not follow robot’s instruc- tions Table 1: Set of possible combinations of request- response pairs (robot and child actions). adaptation, specially in HRI. In this paper we report first results on children adaptive behavior towards a dance tutoring robot. We can observe that children behavior rapidly evolves through few sessions in order to accommodate with the robotic tutor rhythm and instructions. We next briefly describe the dance task and the robotic system used in this work, followed by results obtained from first analysis of the experiments carried out. 2. A DANCE ROBOT TUTOR The present work is based on the approach presented in [5], where creative dance is used as the interaction ac- tivity between robot and child. Creative dance is a form of dance where the goal is to explore the body movement based on a set of general guidelines (movement concepts). Several methods are used to teach dance. In this work, we focus on what we call the concept-based learning. In this model dance concepts are sequentially shown and chil- dren explore the new ideas using their own body. They are free to create any variation they can think of as long as it corresponds to the concept described. Next, a process of request-respond takes place. The robot asks for a dance con- cept and the child has four seconds to perform it. If he/she fails, the robot shows an example and the child may (or not) respond in some way. Table 1 summarizes the available request-respond pairs observed in the experiments we report here.