INTRO: A Multidisciplinary Approach to Intelligent Human-Robot Interaction Aleksandar Jevtić, Eric Lucet Robosoft S.A. Bidart, France {aleksandar.jevtic, eric.lucet}@robosoft.fr Alex Kozlov, Jeremi Gancet Space Applications Services Zaventem, Belgium {alex.kozlov, jeremi.gancet}@spaceapplications.com Abstract— This paper presents the interactive robotics concept being developed by the INTRO research network. The aim is to create a new generation of intelligent mobile robots that operate in close interaction with humans in unstructured, dynamically changing environments. The INTRO network consists of a team of researchers, from academia and industry, which create a multidisciplinary framework that entails Cooperative Robot Learning, Cognitive Human-Robot Interaction (HRI), and Intelligent Interface Design. The robotic system being developed will be tested in two application scenarios: the Robot Waiter, and the Urban Search and Rescue (USAR). For these scenarios, two different robotic platforms are used in the implementation stage. This paper presents an overview of the obtained research objectives, and proposes a framework for the integration of work and the implementation of the expected results. Finally, the paper describes a potential impact through development and use of research results and proposes future lines of research. Keywords – human-robot interaction; learning; human-robot interface; manipulator arm; robot safety; mobile robots; autonomous robots. I. INTRODUCTION Human-Robot Interaction (HRI) is an interdisciplinary study of interaction dynamics between humans and robots. In recent years, this field has received attention from a wider scientific community and it has been greatly supported by both public and private funding. Robots have become part of our everyday lives; they clean our homes, work as restaurant waiters, perform monitoring and surveillance, assist in search and rescue missions, explore outer space, and so forth. These robots are usually designed for a specific mission, and their interaction with humans is limited. They are built to perform simple tasks that do not involve human-robot cooperation. The aim of the INTRO research network is to create a new generation of intelligent mobile robots that operate in close interaction with humans in unstructured, dynamically changing environments. The proposed interactive robotics concept aims to contribute to the areas of Cooperative Robot Learning, Cognitive HRI, and Intelligent Interface Design, and poses a big challenge in terms of integration of results and their implementation on a single robotic platform. The resulting technological developments should provide more intuitive HRI and scalability in terms of a wider range of applications. II. STATE OF THE ART Various scientific disciplines contribute to the field of HRI, which brings collaboration of scientists and engineers of different professional backgrounds. HRI combines research from Computer Science and Engineering, but also human sciences, including cognitive, behavioural, and social sciences [1]. Great research efforts are being made because of high expectations and potentially large impact that HRI may have on quality of living. Robots could replace humans in carrying out work that is physically hard, dangerous, or involves repetitive and boring tasks. Applications in which the cooperation between humans and robots is expected are even greater challenges. HRI is an essential part of robot-assisted applications. The most popular interaction method for hazardous environments is through some form of remote teleoperation or telepresence [2, 3]. This has the obvious advantage of not directly endangering personnel in hazardous areas. The main aim is to achieve remote perception and situational awareness. Another interaction paradigm is direct, proximate interaction and human-robot collaboration (HRC) [4, 5]. In this case interaction capabilities include human gesture recognition, voice recognition and natural language interpretation. Functionalities for useful human-robot collaboration include cognitive spatial reasoning, perspective taking, and behaviours for sequencing and executing high level tasks. Dynamic autonomy is also expected, which would allow the robot to dynamically adjust its behaviours depending on the situation. Natural and efficient human-robot collaboration is crucial to advancing robotic applications. One significant challenge that is common in any form of HRC is synchronization of actions between human and robot [6, 7]. Current HRC is fairly limited to very rigid and structured turn based interaction. Research in timing-based HRC aims to improve the fluency and efficiency of joint HRC making the interactions more natural, coordinated and effective [8]. Joint attention is also an This work has been financed by the EU funded Initial Training Network (ITN) in the Marie-Curie People Programme (FP7): INTRO (INTeractive RObotics research network), grant agreement no.: 238486