Guest Lecture : Cristian Jimenez Romero

Evolving self-organized robotic swarms controlled by Spiking Neural Networks

Cristian Jimenez Romero est spécialisé dans l'informatique, les neurosciences et la robotique. Il est actuellement chercheur post-doctoral Marie Skolodowska/EUTOPIA-SIF au sein de CY Cergy Paris Université et accueilli par le laboratoire ETIS.

My work explores the emergence of self-coordination and communication in evolved swarms, specifically in an ant colony simulation. Unlike conventional approaches that rely on manually defined rules, we use an evolutionary algorithm to optimize a spiking neural network (SNN) for each agent, allowing them to make decisions autonomously.
The objective of the evolved colony is to efficiently forage for food and return it to the nest. During evolution, the ants learn to collaborate by depositing pheromone near food piles and the nest, guiding other ants. Notably, this pheromone-based communication is not manually encoded but arises naturally through optimization.
Our SNN-based model demonstrates a level of foraging efficiency comparable to that of the multi-agent rule-based system.. This work demonstrates that SNNs can facilitate self-coordination and communication in multi-agent systems without the need for predefined rules, opening new possibilities for complex applications using SNN.

Date: 17 October 2023 from 14:00 to 15:30

The hybrid guest lecture is organised in person at the H008 meeting room of MIR in Neuville-sur-Oise and remotely on Zoom.

To attend the remote guest lecture, please connect to Zoom: https://cyu-fr.zoom.us/j/92567729016

Meeting ID: 925 6772 9016

The video will be online on the CY AS YouTube channel