Benefits of Lamarckian Evolution for Morphologically Evolving Robots

Milan Jelisavcic; Rafael Kiesel; Kyrre Glette; Evert Haasdijk; A.E. Eiben

doi:10.1145/3067695.3076046

Benefits of Lamarckian Evolution for Morphologically Evolving Robots

Milan Jelisavcic, Rafael Kiesel, Kyrre Glette, Evert Haasdijk, A.E. Eiben

Research output: Contribution to Conference › Poster › Other research output

125 Downloads (Pure)

Abstract

Implementing lifetime learning by means of on-line evolution, we establish an indirect encoding scheme that combines Compositional Pattern Producing Networks (CPPNs) and Central Pattern Generators (CPGs) as a relevant learner and controller for open-loop gait controllers in modular robots which have evolving morphologies. Experimental validation on the morphologically evolved robots shows that a Lamarckian setup with CPPN-CPG provides substantial benefits compared to controllers learned from scratch.

Original language	English
Number of pages	2
DOIs	https://doi.org/10.1145/3067695.3076046
Publication status	Accepted/In press - Jul 2017
Event	The Genetic and Evolutionary Computation Conference - Germany, Berlin, Germany Duration: 15 Jul 2017 → 19 Jul 2017 Conference number: 18 http://gecco-2017.sigevo.org/index.html/HomePage

Conference

Conference	The Genetic and Evolutionary Computation Conference
Abbreviated title	GECCO 2017
Country/Territory	Germany
City	Berlin
Period	15/07/17 → 19/07/17
Internet address	http://gecco-2017.sigevo.org/index.html/HomePage

Keywords

Evolutionary robotics
On-line evolution
Indirect encoding
Lamarckian evolution
Gait learning

Access to Document

10.1145/3067695.3076046

GECCO2017_abstract_pos213Final published version, 1.41 MB
GECCO2017_posterFinal published version, 1.85 MB

Persistent URL (handle)

Copy link address

Cite this

@conference{9ef2054a26de474a9eefd4903b808383,

title = "Benefits of Lamarckian Evolution for Morphologically Evolving Robots",

abstract = "Implementing lifetime learning by means of on-line evolution, we establish an indirect encoding scheme that combines Compositional Pattern Producing Networks (CPPNs) and Central Pattern Generators (CPGs) as a relevant learner and controller for open-loop gait controllers in modular robots which have evolving morphologies. Experimental validation on the morphologically evolved robots shows that a Lamarckian setup with CPPN-CPG provides substantial benefits compared to controllers learned from scratch.",

keywords = "Evolutionary robotics, On-line evolution, Indirect encoding, Lamarckian evolution, Gait learning",

author = "Milan Jelisavcic and Rafael Kiesel and Kyrre Glette and Evert Haasdijk and A.E. Eiben",

year = "2017",

month = jul,

doi = "10.1145/3067695.3076046",

language = "English",

note = "The Genetic and Evolutionary Computation Conference, GECCO 2017 ; Conference date: 15-07-2017 Through 19-07-2017",

url = "http://gecco-2017.sigevo.org/index.html/HomePage",

}

TY - CONF

T1 - Benefits of Lamarckian Evolution for Morphologically Evolving Robots

AU - Jelisavcic, Milan

AU - Kiesel, Rafael

AU - Glette, Kyrre

AU - Haasdijk, Evert

AU - Eiben, A.E.

N1 - Conference code: 18

PY - 2017/7

Y1 - 2017/7

N2 - Implementing lifetime learning by means of on-line evolution, we establish an indirect encoding scheme that combines Compositional Pattern Producing Networks (CPPNs) and Central Pattern Generators (CPGs) as a relevant learner and controller for open-loop gait controllers in modular robots which have evolving morphologies. Experimental validation on the morphologically evolved robots shows that a Lamarckian setup with CPPN-CPG provides substantial benefits compared to controllers learned from scratch.

AB - Implementing lifetime learning by means of on-line evolution, we establish an indirect encoding scheme that combines Compositional Pattern Producing Networks (CPPNs) and Central Pattern Generators (CPGs) as a relevant learner and controller for open-loop gait controllers in modular robots which have evolving morphologies. Experimental validation on the morphologically evolved robots shows that a Lamarckian setup with CPPN-CPG provides substantial benefits compared to controllers learned from scratch.

KW - Evolutionary robotics

KW - On-line evolution

KW - Indirect encoding

KW - Lamarckian evolution

KW - Gait learning

U2 - 10.1145/3067695.3076046

DO - 10.1145/3067695.3076046

M3 - Poster

T2 - The Genetic and Evolutionary Computation Conference

Y2 - 15 July 2017 through 19 July 2017

ER -

Benefits of Lamarckian Evolution for Morphologically Evolving Robots

Abstract

Conference

Keywords

Access to Document

Persistent URL (handle)

Fingerprint

Cite this