Impact of Neuron Models and Network Structure on Evolving Modular Robot Neural Network Controllers

Léo Cazenille 1, 2 Nicolas Bredeche 1, 2, * Heiko Hamann 3 Jürgen Stradner 3
* Corresponding author
2 TAO - Machine Learning and Optimisation
CNRS - Centre National de la Recherche Scientifique : UMR8623, Inria Saclay - Ile de France, UP11 - Université Paris-Sud - Paris 11, LRI - Laboratoire de Recherche en Informatique
Abstract : This paper investigates the properties required to evolve Artificial Neural Networks for distributed control in mod- ular robotics, which typically involves non-linear dynamics and complex interactions in the sensori-motor space. We in- vestigate the relation between macro-scale properties (such as modularity and regularity) and micro-scale properties in Neural Network controllers. We show how neurons capable of multiplicative-like arithmetic operations may increase the performance of controllers in several ways whenever chal- lenging control problems with non-linear dynamics are in- volved. This paper provides evidence that performance and robustness of evolved controllers can be improved by a com- bination of carefully chosen micro- and macro-scale neural network properties.
Document type :
Conference papers
Complete list of metadatas

https://hal.inria.fr/hal-00731411
Contributor : Nicolas Bredeche <>
Submitted on : Wednesday, September 12, 2012 - 4:43:20 PM
Last modification on : Thursday, April 5, 2018 - 12:30:12 PM
Long-term archiving on : Friday, December 16, 2016 - 12:36:29 PM

File

2012-GECCO-multipole.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00731411, version 1

Collections

Citation

Léo Cazenille, Nicolas Bredeche, Heiko Hamann, Jürgen Stradner. Impact of Neuron Models and Network Structure on Evolving Modular Robot Neural Network Controllers. GECCO - Genetic and Evolutionary Computation Conference, 2012, Philadelphia, United States. pp.89-96. ⟨hal-00731411⟩

Share

Metrics

Record views

887

Files downloads

416