Context
In a context where demand for increasingly fast, energy-efficient intelligent systems capable of processing complex temporal data continues to grow, spiking neural networks (SNNs) are emerging as a major technology. Inspired by the biological functioning of the brain, SNNs process information in the form of discrete events—spikes rather than using continuous values like traditional neural networks.
Objectives
This internship is part of a research program aimed at developing a new end-to-end autonomous driving architecture adapted to specific missions. It should enable the first steps to be taken toward etablishing SNN networks dedicated to decision-making in the case of overtaking maneuvers. The main objectives are :
(1) through an exhaustive state-of-the-art review, to propose a decision-making method for triggering the maneuver,
(2) to implement the method on a GPU or neuromorphic system by interfacing it with a road traffic simulator (CARLA),
(3) and finally, to propose a benchmark for comparing the proposed solution with traditional ANN-type solutions.
Where ?
IRIMAS UR 7499, Universite de Haute-Alsace (UHA), Mulhouse
Contacts
Benoit Vigne & Jonathan Ledy
@ benoit.vigne@uha.fr & @ jonathan.ledy@uha.fr
