My research interests are in the areas of computer and robot vision, to address two questions: (i) how to develop artificial (robotic) systems that use vision and other sensing modalities to "understand" the world around them and (ii) to investigate, model and understand how vision is used in biological systems like insects or humans. One example is vision based-navigation, motion perception and control, including how to use video imagery in closed loop for mapping and control of land, air and underwater vehicles. It often requires the design of task-specific camera geometries, like omnidirectional and space-variant cameras. This research draws inspiration from biology where evolution has lead to specialized and sophisticated solutions (both eye design and processing) adapted to the behaviour of the various species (e.g. compound eyes, foveal cameras). Another main area of research is cognitive humanoid systems. Together with neuroscientists and developmental psychologists, we aim to understand human cognition, sensorimotor coordination and learning, as well as to design robotic systems with learning and adaptation capabilities. Flagship projects have addressed the models of mirror neurons to enhance the ability of human action recognition, imitation and programming by demonstration. We have proposed a model of object affordances that improves a robot’s ability to manipulate objects, ground language and choose tools that are adequate to a given function. I have co-developed several advanced robotic platforms, such as the humanoid robot Baltazar, the social robot Vizzy and the iCub, the most advanced open-source humanoid robot in the world. These systems integrate the RBCog-Lab, that is part of the Portuguese Roadmap for Research Infrastructures, in Robotics, Brain and Cognition.