As part of my PhD, I created a spiking neural network capable of making directional decisions which leads a robot along paths of different shapes in a real-world environment. The robot performs fully autonomously, using only locally sourced olfactory data and no additional sensors. The SNN is in full control of the robot’s navigational decisions. This SNN was developed to infer the direction of a stimulus using a stereo-olfaction approach. It integrates the efficient and accurate slope-detector Izhikevich points neurons (Miko et al. 2024). Confronting live data in a real-world environment and facing noise-ridden, turbulent-induced signals tested the success of this approach.

This post is a part of the supplementary material for my thesis. If you are interested in seeing more details, I will soon provide a reference to the published preprint.

References:

Miko, Rebecca, Marcus Scheunemann, Volker Steuber, and Michael Schmuker (2024). “Decoding the amplitude and slope of continuous signals into spikes with a spiking point neuron model”. In: Preprint. doi: 10.1101/2024.05.20.594931v1.

Funding:

Funded by the 6G-Life project and was conducted at the Chair of Material Science and Nanotechnology.