Modular genetic neural networks for 6-Legged locomotion

Gruau, Frédéric

doi:10.1007/3-540-61108-8_39

Cited by 9 publications

(11 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, relevant work includes Koza (1995), who has used gene duplication in genetic programming, and Gruau (1995), who has proposed a genetic encoding scheme for neural networks based on cellular duplication and differentiation process. For an interesting discussion on how gene duplication supports modularity see, also, Rotaru-Varga (1999).…”

Section: Introductionmentioning

confidence: 99%

Duplication of Modules Facilitates the Evolution of Functional Specialization

Calabretta

Nolfi²,

Parisi³

et al. 2000

Artificial Life

View full text Add to dashboard Cite

The evolution of simulated robots with three different architectures is studied. We compared a non-modular feed-forward network, a hardwired modular and a duplication-based modular motor control network. We conclude that both modular architectures outperform the non-modular architecture, both in terms of rate of adaptation as well as the level of adaptation achieved. The main difference between the hardwired and duplication-based modular architectures is that in the latter the modules reached a much higher degree of functional specialization of their motor control units with regard to high level behavioral functions. The hardwired architectures reach the same level of performance, but have a more distributed assignment of functional tasks to the motor control units. We conclude that the mechanism through which functional specialization is achieved is similar to the mechanism proposed for the evolution of duplicated genes. It is found that the duplication of multifunctional modules first leads to a change in the regulation of the module, leading to a differentiation of the functional context in which the module is used. Then the module adapts to the new functional context. After this second step the system is locked into a functionally specialized state. We suggest that functional specialization may be an evolutionary absorption state.

show abstract

Section: Introductionmentioning

confidence: 99%

Duplication of Modules Facilitates the Evolution of Functional Specialization

Calabretta

Nolfi²,

Parisi³

et al. 2000

Artificial Life

View full text Add to dashboard Cite

show abstract

“…The goal of the performance simulator is to validate the concept of "particles" and show they can be used to implement complex computations. Due to the spatial nature of Blob executions, the simulator comes with a graphical user interface for visualizing the development of Blob programs, see Figures 19,23. In the next paragraphs, we show how to program three fairly different kinds of problems using a Blob computer: the QuickSort algorithm, the Kung and Leiserson [37] matrix multiply algorithm, and a neural network program used for robot control [27]. Performance metrics are shown for the QuickSort algorithm only.…”

Section: Experimental Frameworkmentioning

confidence: 99%

BLOB computing

Gruau

Lhuillier

Reitz

et al. 2004

Proceedings of the 1st Conference on Computing Frontiers

Self Cite

View full text Add to dashboard Cite

Current processor and multiprocessor architectures are almost all based on the Von Neumann paradigm. Based on this paradigm, one can build a general-purpose computer using very few transistors, e.g., 2250 transistors in the first Intel 4004 microprocessor. In other terms, the notion that on-chip space is a scarce resource is at the root of this paradigm which trades on-chip space for program execution time. Today, technology considerably relaxed this space constraint. Still, few research works question this paradigm as the most adequate basis for high-performance computers, even though the paradigm was not initially designed to scale with technology and space.In this article, we propose a different computing model, defining both an architecture and a language, that is intrinsically designed to exploit space; we then investigate the implementation issues of a computer based on this model, and we provide simulation results for small programs and a simplified architecture as a first proof of concept. Through this model, we also want to outline that revisiting some of the principles of today's computing paradigm has the potential of overcoming major limitations of current architectures.

show abstract

“…For example, early works with Cellular Encoding implemented regularity through recursive decomposition, and was successfully applied to a six-legged robot locomotion problem [5]. Recent works focused on how such properties emerge during evolution, either with direct encoding [11,2] or indirect encoding approaches [22].…”

Section: Network Structurementioning

confidence: 99%

Impact of neuron models and network structure on evolving modular robot neural network controllers

Cazenille

Bredèche

Hamann

et al. 2012

Proceedings of the 14th Annual Conference on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

This paper investigates the properties required to evolve Artificial Neural Networks for distributed control in modular robotics, which typically involves non-linear dynamics and complex interactions in the sensori-motor space. We investigate 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 challenging control problems with non-linear dynamics are involved. This paper provides evidence that performance and robustness of evolved controllers can be improved by a combination of carefully chosen micro-and macro-scale neural network properties.

show abstract

Modular genetic neural networks for 6-Legged locomotion

Cited by 9 publications

References 9 publications

Duplication of Modules Facilitates the Evolution of Functional Specialization

Duplication of Modules Facilitates the Evolution of Functional Specialization

BLOB computing

Impact of neuron models and network structure on evolving modular robot neural network controllers

Contact Info

Product

Resources

About