Dipartimento di Ingegneria informatica, automatica e gestionale

Human-Robot Interaction (HRI) has become increasingly important as robots have been integrated into various daily life aspects. Gesture recognition plays a crucial role in non-verbal communication in HRI. Indeed, the real-time robot's interpretation of human gestures is essential for enhancing the overall user experience. To this aim, improving the accuracy and efficiency of gesture recognition in HRI could be achieved by integrating Virtual Reality (VR) technology. In this article, we describe the development of virtually-generated avatars and a gesture recognition system that uses machine learning techniques to allow ground robots to recognize gestures performed by both digital agents and actual humans. Nevertheless, we address some existing challenges by presenting a set of gesture definitions, data generation, and system evaluation using virtual reality simulation. Our findings and results highlight the importance of addressing these challenges to enable effective gesture recognition in HRI. In particular, we demonstrate that the adoption of VR can significantly increase the system's accuracy and efficiency, improving the overall user experience.