Evolution of Neural Controllers with Adaptive Synapses and Compact Genetic Encoding

Floreano, Dario; Urzelai, Joseba

doi:10.1007/3-540-48304-7_25

Cited by 10 publications

(9 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…When tested in gray and black environments, only a few lucky individuals that encounter the target before a wall have non-zero tness values. Individuals with noisy synapses score very low tness values in all conditions but they generalize better than genetically-determined individuals 6 . 4 Using di erent sequences of random number.…”

Section: Resultsmentioning

confidence: 96%

See 1 more Smart Citation

Evolutionary Robots with Fast Adaptive Behavior in New Environments

Urzelai

Floreano

2000

Evolvable Systems: From Biology to Hardware

Self Cite

View full text Add to dashboard Cite

Abstract. This paper is concerned with adaptation capabilities of evolved neural controllers. A method consisting of encoding a set of local adaptation rules that synapses obey while the robot freely moves in the environment 6] is compared to a standard xed-weight network. In the experiments presented here, the performance of the robot is measured in environments that are di erent in signi cant w ays from those used during evolution. The results show that evolutionary adaptive controllers can adapt to environmental changes that involve new sensory characteristics (including transfers from simulation to reality) and new spatial relationships. Evolution and AdaptationEvolutionary algorithms are widely used in autonomous robotics in order to solve a large variety of tasks in several kind of environments. However, evolved controllers become well adapted to environmental conditions used during evolution, but often do not perform well when conditions are changed. Under these circumstances, it is necessary to carry on the evolutionary process, but this might t a k e long time.Combination of evolution and learning has been shown to be a viable solution to this problem by providing richer adaptive dynamics 1] than in the case where parameters are entirely genetically-determined. A review of the work combining evolution and learning for sensory-motor controllers can be found in 5, 9].Instead of simply combining o -the-shelf evolutionary and learning algorithms, in previous work we presented an approach capable of generating adaptive neural controllers by evolving a set of simple adaptation rules 6]. The method consists of encoding on the genotype a set of modi cation rules that perform Hebbian synaptic changes 2{4] through the whole individual's life. The results showed that evolution of adaptive individuals generated viable controllers in much less generations and that these individuals displayed more performant behaviors than genetically-determined individuals.In this paper, we describe two new sets of experiments conceived to measure the adaptation capabilities of this approach i n environments that are di erent from those used during evolution. The results are compared to standard evolution of synaptic weights and to evolution of noisy synaptic weights (control condition).

show abstract

Section: Resultsmentioning

confidence: 96%

“…5 Node Encoding for xed synapses was not capable of solving the original problem, therefore we report results for Synapse Encoding. 6 Notice that adaptive individuals report better tness also in the evolutionary environment. The performance issue has been addressed in another paper 6].…”

Section: Resultsmentioning

confidence: 99%

Evolutionary Robots with Fast Adaptive Behavior in New Environments

Urzelai

Floreano

2000

Evolvable Systems: From Biology to Hardware

Self Cite

View full text Add to dashboard Cite

show abstract

“…Ainsi, comme (Floreano et al, 1999(Floreano et al, , 2000 on peut proposer différentes lois locales d'adaptation des poids inspirées de la biologie : règle de Hebb, règle post-synaptique, règle pré-synaptique, règle de covariance... Nous proposons dans notre démarche d'incorporer en plus d'autres lois biologiques de type « homéostatique » (Turrigiano, 1999) qui vont tendre par synergie à stabiliser la fréquence d'activation du neurone dans un domaine efficace. Ainsi, un neurone « saturé » ou « muet » verra son excitabilité (son biais) diminuer ou augmenter vers un niveau d'activité raisonnable.…”

Section: Modèle Bio-inspiré De Contrôleur Neuronal Adaptatifunclassified

Étude sur l'intérêt des modèles biologiques de réseaux de neurones pour la synthèse de rythmes locomoteurs adaptatifs

Hoinville¹,

Hénaff²,

Delaplace³

2007

ournal Européen des Systèmes Automatisés

View full text Add to dashboard Cite

show abstract

“…[5] in which a local adaptation rule is assignated to each connection as suggested by biological observations: In order to model the mechanism that regulate neuronal excitability, we use the model of center crossing networks proposed by Mathayomchan and al. [13].…”

Section: B Neuron and Synapse Modelmentioning

confidence: 99%

“…We adapt this concept to build our versatile model by reformulating it. Thus, after transformation 4 and homogenization 5 , we obtain the modified activity o i as follows:…”

Section: B Neuron and Synapse Modelmentioning

confidence: 99%

Comparative study of two homeostatic mechanisms in evolved neural controllers for legged locomotion

Hoinville

Hénaff

2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE Cat. No.04CH37566)

View full text Add to dashboard Cite

Abstract-This paper presents a preliminary study on the advantages of two bio-inspired homeostatic mechanisms in neural controllers of legged robots. We consider a robot made up of one leg of 3 dof pushing a body that is sliding on a rail with a friction force. The synthesis of the controller is done by an evolutionary algorithm which choose to attach to each synapse a particular plastic law. Four models of network incorporating or not each homeostatic law are proposed. After evolution, effectiveness of each kind of adaptive controllers is compared in term of statistics on a task of controlling the speed of the robot. The robustness to a perturbation generated by the viscous friction is analyzed in term of control. Results show that homeostatic mechanisms increase evolvability, stability and adaptivity of those controllers.

show abstract

Evolution of Neural Controllers with Adaptive Synapses and Compact Genetic Encoding

Cited by 10 publications

References 5 publications

Evolutionary Robots with Fast Adaptive Behavior in New Environments

Evolutionary Robots with Fast Adaptive Behavior in New Environments

Étude sur l'intérêt des modèles biologiques de réseaux de neurones pour la synthèse de rythmes locomoteurs adaptatifs

Comparative study of two homeostatic mechanisms in evolved neural controllers for legged locomotion

Contact Info

Product

Resources

About