Future robots, instead, should be able to learn how to truly “master” their environments autonomously, i.e. to self-generate goals and autonomously and efficiently learn the skills to accomplish them based on the progressive acquisition, modification, generalisation, and recombination of previously learned skills and knowledge. This will allow them, with little additional learning, to change an environment from its current state to a wide range of potential goal states desired by the user. The question is: how can we create future robots that are able to face this challenge?
The GOAL-Robots project
Addressing this question and having central importance for applications and artificial intelligence, is the start of a new European project led by the Laboratory of Computational Embodied Neuroscience (LOCEN), an Italian research group based in Rome at the Institute of Cognitive Sciences and Technologies of the Italian National Research Council (ISTC-CNR).
“GOAL-Robots – Goal-based Open-ended Autonomous Learning Robots” project ranked #1 among all 11 projects funded out of the 800 participants to the April 2016 EU FET-OPEN call (Future Emergent Technologies), and is part of the Horizon 2020 EU research program. LOCEN and its Principal Investigator, Gianluca Baldassarre, will coordinate the project consortium also involving other three important European research groups:
the Laboratoire Psychologie de la Perception (LPP, France), headed by Kevin O’Regan and based in Paris at the Institut Paris Descartes de Neurosciences et Cognition, that will carry out experiments on the learning of goals and skills in children;
the Frankfurt Institute of Advanced Studies (FIAS, Germany), headed by Jochen Triesch, focussed on the development of bio-inspired visual and attention systems and motor synergies; and
the group of roboticists directed by Jan Peters, based at the Technische Universitaet Darmstadt (TUDa, Germany), which will lead the challenging robotic demonstrators of the project.