Elisa Massi and Lorenzo Vannucci

Speakers

Elisa Massi is a Ph.D. student in Computational Neuroscience and Robotics at the Institute of Intelligent Systems and Robotics (ISIR) at Sorbonne University. She is advised by Doctors Benoît Girard and Mehdi Khamassi and she is interested in integrating brain-like abilities and skills in artificial agents and robots. Her current research focuses on data-driven behavioural modelling of spatial exploration and on studying the role of hippocampal replay in navigation and learning, to transfer the principle from neuroscience to Reinforcement Learning and eventually to Robotics. Elisa received a Bachelor of Science in Automation Engineering at the University of Bologna, in collaboration with the Tongji University of Shaghai in 2015, and a Master of Science in Bionics Engineering at Sant’Anna School of Advanced Studies in Pisa in 2018, with a master thesis in collaboration with the Technical University of Denmark (DTU), entitled “Design and implementation of a bio-inspired control architecture for locomotion”.

Lorenzo Vannucci is a Post-Doctoral fellow at the Institut de la Vision, Soronne Université - CNRS - INSERM. He obtained his Ph.D. in BioRobotics from Scuola Superiore Sant’Anna in 2018 with a thesis entitled “Brain-inspired methods for adaptive and predictive control of humanoid robots” and his M.Sc. in Computer Science at the University of Pisa in 2014 with a thesis entitled “An adaptive neuro-controller for head-stabilized visual pursuit in a humanoid robot”. His current research activity involves the development of event-based algorithms for sensory processing.

Abstract

Animals are able to negotiate complex environments not only thanks to their advanced bodies, but most importantly to the orchestration performed by the nervous system, and especially by the brain. The main feature of the brain is being capable to process a massive amount of sensory information received and to use it to perform both cognitive and motor tasks, often at the same time. The aim of neurorobotics is to exploit such incredible feature by creating computational models that replicates them and embedding such models in robotics controllers. This can in principle gives robots capabilities such as predicting changes in the environment and reacting accordingly, perform a careful motion planning based on energy/rewards constraints, naturally process sensory inputs and execute animal-like motions. In this presentation we will present two examples of neurorobotics in action. In the first one, we will show how data-driven models of the brain mechanisms that orchestrate hippocampal replay can be used, together with a reinforcement learning system, to create efficient and adaptive robotic navigation. In the second example, we will see how to process event-based sensors with a biologically inspired sensory pipeline.