neurocybernetic team

Home | Publications | Members | Internships | Medias | Videos | Access | Agenda


Neurocybernetic team

Responsable : Alexandre Pitti

Developmental and Bio-inspired Robotics

The work of the Neurocybernetic team follows 4 main axes:
  • Neurobiological modeling of navigation and meta-control mechanisms,
  • the proposal of models for sensorimotor development,
  • the study of the emergence of social interactions
  • active perception and multimodality

    The goal is to build precise computational models of the brain : relationship between associative brain areas, hippocampus, basal ganglia, cerebelum and prefrontal cortex. Models are implemented and tested on robots in real environements with the double purpose of :

    • Verifying the hypothesis done on the biological models and learning from failures and impossibilities in order to correct and complete those models. This is done by the mean of many feedbacks with psychologists and neuroscientists.
    • Improving the behaviors of the robots in order to make them more autonomous with a better integration with humans (robustness, interaction). The goal is to directly apply what has been learned and discover in concret and useful applications.

    Neurobiological modeling of navigation and meta-control mechanisms

    Our modeling work aims on the one hand to improve our understanding of the neural mechanisms implemented by the mammalian brain when it is engaged in a navigation task and on the other hand to propose a neural architecture allowing a robot to navigate in real environments by exploiting different sensorimotor strategies.

    More information: Autonomous navigation

    Studying the emergence of social interactions.

    More information: Human-robot interactions

    The sensorimotricity of our body structures the way we perceive the world.

    Our perception of the environment depends a lot on how we move and where our sensory receptors are. The same goes for robots. The geometry of the robot (its morphology), the composition of its materials (its visco-elasticity) the arrangement of its sensors and its motors structure the information that arrives to it. This is called sensorimotor coordination or the perception-action loop.

    The self-organization of brain structures emerges from development and learning.

    As for a child, a robot can learn this sensorimotor coupling, to control the actions of its own body and to actively represent the external space through its sensors, by making as for an infant a babbling which will structure its system of representations internal (his memory) and his sensorimotor controller: in short, his brain.

    Research Master

    The team is also associated with the Research master "artificial intelligence and robotics" of the University of Cergy-Pontoise.

    TEDx presentation 2016

    A. Pitti

    Overview of the lab

  • © 2011–2018 — ETIS UMR 8051 — webmaster Arnaud Blanchard