2026-04-14T08:31:21Zhttps://recercat.cat/oai/request

oai:recercat.cat:2117/3906442026-01-20T12:25:34Zcom_2072_1033col_2072_452951

Learning new human gestures through Human Robot Interaction Grifols Vilella, Lluis Universitat Politècnica de Catalunya. Departament d'Enginyeria de Sistemes, Automàtica i Informàtica Industrial Garrell Zulueta, Anais Pino Bastida, Iván del Àrees temàtiques de la UPC::Informàtica::Robòtica Human-robot interaction Computer vision Human-Robot Interaction ROS In Computer Vision Hand Gesture Recognition Interacció Humà-Robot ROS Intel·ligència artificial Visió artificial Reconeixement de Gestos Interacció persona-robot Visió per ordinador Hand gesture recognition has been extensively used for human-robot interaction in recent years, as they provide a simple and intuitive approach to communication with social and assistive robots. For many real-world applications, it is desirable to have robots able to learn new concepts through interaction, however, this is not a trivial problem: artificial neural networks --commonly used for hand gesture recognition-- suffer from catastrophic forgetting, which hinders their capability of learning new gestures. To cope with this problem, we develop a system for incremental learning of hand gestures applied to human-robot interaction. This system is deployed and tested in a real robot demonstrating the ability to learn and identify correctly most of the new hand gestures examples shown. 2023-05-16 Master thesis https://hdl.handle.net/2117/390644 173403 eng Restricted access - author's decision application/pdf Universitat Politècnica de Catalunya