Biography

Stephane Doncieux is Professor in Computer Science at ISIR (Institute of Intelligent Systems and Robotics), Sorbonne University, CNRS, in Paris, France. Since January 2018, he is deputy director of the ISIR, a multidisciplinary robotics laboratory with researchers in mechatronics, signal processing computer science and neuroscience. Until that date, he was in charge of the AMAC multidisciplinary research team (Architectures and Models of Adaptation and Cognition). He was coordinator of the DREAM FET H2020 project from 2015 to 2018 (http://robotsthatdream.eu/). His research is in cognitive robotics, with a focus on learning and adaptation with a developmental approach.

Abstract

Tentative title: Open-ended learning in robotics

Reinforcement learning is a reward-driven adaptation process. This formalism has been very successful in defining learning processes in both robotics and neuroscience. However, it requires the definition of a state space and an action space that the learning will explore. These state and action spaces determine the type of task that the robot can perform. Moreover this definition requires a significant expertise and the performance of the learning process is strongly dependent on this choice. The remarkable human cognitive abilities are, according to psychologists like A. Karmiloff-Smith, related to our capacity to rewrite our knowledge in a different format, potentially more suited to solving the tasks we face. The European project DREAM aimed to provide robots with such a capacity by focusing on the construction and redescription of state and action spaces. The project focused in particular on the bootstrap phase of this process: the creation of state spaces and actions starting from very limited knowledge about the environment and its characteristics. The approach developed during the project will be presented, as well as the results that have been obtained.