Project description DEENESFRITPL Adaptive self-aware robots Robotics is increasingly focusing on autonomous adaptive and learning systems that can interact with people in a predictable and intuitive manner. Currently, these technologies depend on the explicit design of robots using pre-engineered models. To become independent and adapt quickly to situations unforeseen by their creators, it's crucial for robots to learn to anticipate the sensory consequences of intended actions. The EU-funded Predictive Robots project addresses this challenge by developing techniques for curiosity-driven exploration behaviours that are typical of infant development. Conducted on humanoid robots and marine drones, the project also envisions creating an artificial self with perceptual skills and subjective experiences. Show the project objective Hide the project objective Objective Current robot technologies are still not enough autonomous and adaptive to safely and intuitively interact with people. Nowadays, artificial systems mostly rely on pre-engineered models of the world and of their embodiment. Defining such models a-priori can be very challenging and may result in robots lacking the capability to react to situations that are not foreseen by their designers. The forthcoming societal needs and economic opportunities for robotics require smarter, more adaptive and self-aware artificial systems. This project addresses this challenge by developing new methods: (1) for the autonomous acquisition of models of the robot’s body inspired on infant development, where online deep learning techniques are integrated within curiosity-driven exploration strategies for high-dimensional task spaces; (2) for the enhancement of robots’ perceptual skills based on predictive processes, such as visual input enhancement through the attenuation of expected perceptions of self-body movements; (3) for studying possibilities of an artificial Self and of providing subjective experiences to robots. The core of the proposed research lays on predictive models learned through exploration behaviours typical of infants’ development. Predictive robots will be capable of anticipating sensory consequences of intended actions. This research has a broad range of applications - such as low-cost improvement of current sensing technologies - and can advance the understanding of brain processes behind particular phenomena - such as the sense of object permanence, memory, self-awareness and sense of agency - beside providing insights for their implementation into artificial systems. The research project will be conducted on humanoid robots and marine drones at the BioRobotics Institute of the Scuola Superiore di Studi Universitari e di Perfezionamento Sant'Anna in Pisa, Italy. Fields of science engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringroboticsautonomous robotsdronesnatural sciencescomputer and information sciencesartificial intelligencemachine learningdeep learning Programme(s) H2020-EU.1.3. - EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions Main Programme H2020-EU.1.3.2. - Nurturing excellence by means of cross-border and cross-sector mobility Topic(s) MSCA-IF-2018 - Individual Fellowships Call for proposal H2020-MSCA-IF-2018 See other projects for this call Funding Scheme MSCA-IF-EF-ST - Standard EF Coordinator SCUOLA SUPERIORE DI STUDI UNIVERSITARI E DI PERFEZIONAMENTO S ANNA Net EU contribution € 183 473,28 Address PIAZZA MARTIRI DELLA LIBERTA 33 56127 Pisa Italy See on map Region Centro (IT) Toscana Pisa Activity type Higher or Secondary Education Establishments Links Contact the organisation Opens in new window Website Opens in new window Participation in EU R&I programmes Opens in new window HORIZON collaboration network Opens in new window Total cost € 183 473,28
