Department of Computer, Control and Management Engineering

Recent advancements in robotics have driven the development of autonomous systems capable of navigating and interacting with complex environments. This work focuses on enhancing robotic autonomy by building richer representations of the surroundings and enabling robots to adapt their behavior dynamically. By integrating a broad spectrum of sensory inputs with semantic information, three key elements are highlighted: constructing comprehensive environment representations, layering semantic understanding for context-aware decision-making, and designing benchmarking scenarios to evaluate environmental awareness. Leveraging machine learning and model-based techniques, the focus of the seminar will be on improvements in navigation, interaction, and autonomy across both single and multi-robot systems, including humanoids. Real-world validations, such as RoboCup and SciRoc competitions, showcase the effectiveness of these approaches, supported by novel metrics for evaluating semantic understanding. The presented works propose methodologies that empower robots to perceive, reason, and act intelligently in shared human-robot environments.