Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments
This article investigates the neural network system in the leech heartbeat and develops a real-time biomimetic digital hardware using few-resource implementation for hybrid experiments. The leech heartbeat neural network is one of the most simple central pattern generators (CPG). In biology, CPG pro...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2013-11-01
|
Series: | Frontiers in Neuroscience |
Subjects: | |
Online Access: | http://journal.frontiersin.org/Journal/10.3389/fnins.2013.00215/full |
id |
doaj-b0b01f2c83fb4e0a9ff8050af144de45 |
---|---|
record_format |
Article |
spelling |
doaj-b0b01f2c83fb4e0a9ff8050af144de452020-11-24T21:06:49ZengFrontiers Media S.A.Frontiers in Neuroscience1662-453X2013-11-01710.3389/fnins.2013.0021563391Real-time biomimetic Central Pattern Generators into FPGA for hybrid experimentsMatthieu eAmbroise0Timothée eLevi1Sébastien eJoucla2Blaise eYvert3Sylvain eSaïghi4Laboratoire IMS, Université Bordeaux 1Laboratoire IMS, Université Bordeaux 1Laboratoire INCIA, University of BordeauxLaboratoire INCIA, University of BordeauxLaboratoire IMS, Université Bordeaux 1This article investigates the neural network system in the leech heartbeat and develops a real-time biomimetic digital hardware using few-resource implementation for hybrid experiments. The leech heartbeat neural network is one of the most simple central pattern generators (CPG). In biology, CPG provide for rhythmic bursts of spikes and is the basis for all muscles contractions orders (heartbeat) and locomotion (walking, running….). The leech neural network system was already investigated and this CPG has been already formalized with Hodgkin-Huxley neural model (HH) that is the most complex neuron model. However, the resources needed for a neural model is proportional to its complexity. To answer to this issue, this article describes a biomimetic implementation into FPGA (Field Programmable Gate Array) of a network of 240 CPGs using a simple model (Izhikevich model) and by proposing a new synapse model: activity dependent depression synapse. The architecture of the network implementation allows working on a single computation core. This digital system works in real-time, needs few resources and has the same bursting activity behavior than complex model. To validate our implementation of this CPG, we compare it firstly with a simulation of the complex model. Then we match its activity with the pharmacological data of the activity of the rat’s spinal cord. This digital system allows future hybrid experiments and will be a great step towards hybridation between biological tissue and artificial neural network. This network of CPG could be also useful for mimic the activity of a different animal locomotion or developing hybrid experiments for neuroprosthesis development.http://journal.frontiersin.org/Journal/10.3389/fnins.2013.00215/fullcentral pattern generatorneuron modelBiomimeticFPGAspiking neural networksdigital hardware |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Matthieu eAmbroise Timothée eLevi Sébastien eJoucla Blaise eYvert Sylvain eSaïghi |
spellingShingle |
Matthieu eAmbroise Timothée eLevi Sébastien eJoucla Blaise eYvert Sylvain eSaïghi Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments Frontiers in Neuroscience central pattern generator neuron model Biomimetic FPGA spiking neural networks digital hardware |
author_facet |
Matthieu eAmbroise Timothée eLevi Sébastien eJoucla Blaise eYvert Sylvain eSaïghi |
author_sort |
Matthieu eAmbroise |
title |
Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments |
title_short |
Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments |
title_full |
Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments |
title_fullStr |
Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments |
title_full_unstemmed |
Real-time biomimetic Central Pattern Generators into FPGA for hybrid experiments |
title_sort |
real-time biomimetic central pattern generators into fpga for hybrid experiments |
publisher |
Frontiers Media S.A. |
series |
Frontiers in Neuroscience |
issn |
1662-453X |
publishDate |
2013-11-01 |
description |
This article investigates the neural network system in the leech heartbeat and develops a real-time biomimetic digital hardware using few-resource implementation for hybrid experiments. The leech heartbeat neural network is one of the most simple central pattern generators (CPG). In biology, CPG provide for rhythmic bursts of spikes and is the basis for all muscles contractions orders (heartbeat) and locomotion (walking, running….). The leech neural network system was already investigated and this CPG has been already formalized with Hodgkin-Huxley neural model (HH) that is the most complex neuron model. However, the resources needed for a neural model is proportional to its complexity. To answer to this issue, this article describes a biomimetic implementation into FPGA (Field Programmable Gate Array) of a network of 240 CPGs using a simple model (Izhikevich model) and by proposing a new synapse model: activity dependent depression synapse. The architecture of the network implementation allows working on a single computation core. This digital system works in real-time, needs few resources and has the same bursting activity behavior than complex model. To validate our implementation of this CPG, we compare it firstly with a simulation of the complex model. Then we match its activity with the pharmacological data of the activity of the rat’s spinal cord. This digital system allows future hybrid experiments and will be a great step towards hybridation between biological tissue and artificial neural network. This network of CPG could be also useful for mimic the activity of a different animal locomotion or developing hybrid experiments for neuroprosthesis development. |
topic |
central pattern generator neuron model Biomimetic FPGA spiking neural networks digital hardware |
url |
http://journal.frontiersin.org/Journal/10.3389/fnins.2013.00215/full |
work_keys_str_mv |
AT matthieueambroise realtimebiomimeticcentralpatterngeneratorsintofpgaforhybridexperiments AT timotheeelevi realtimebiomimeticcentralpatterngeneratorsintofpgaforhybridexperiments AT sebastienejoucla realtimebiomimeticcentralpatterngeneratorsintofpgaforhybridexperiments AT blaiseeyvert realtimebiomimeticcentralpatterngeneratorsintofpgaforhybridexperiments AT sylvainesaighi realtimebiomimeticcentralpatterngeneratorsintofpgaforhybridexperiments |
_version_ |
1716764604061712384 |