Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Creating high-level components with a generative representation for body-brain evolution.

Gregory S Hornby1, Jordan B Pollack

  • 1DEMO Lab, Computer Science Department, Brandeis University, Waltham, MA 02454-9110, USA.

Artificial Life
|January 23, 2003
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A unicellular walker controlled by a microtubule-based finite-state machine.

Current biology : CB·2022
Same author

Computer-automated evolution of an X-band antenna for NASA's Space Technology 5 mission.

Evolutionary computation·2010
Same author

Computer science. Toward a smarter Web.

Science (New York, N.Y.)·2009
Same author

Evolutionary dynamics of finite populations in games with polymorphic fitness equilibria.

Journal of theoretical biology·2007
Same author

Modular interdependency in complex dynamical systems.

Artificial life·2005
Same author

Ideal evaluation from coevolution.

Evolutionary computation·2004
Same journal

If Turing Played Piano With an Artificial Partner.

Artificial life·2026
Same journal

Discovering Partial Differential Equations With Neural Cellular Automata.

Artificial life·2026
Same journal

Book Review: Exploring the Boundaries of Life-as-It-Is.

Artificial life·2026
Same journal

System 0/1/2/3: Quad-Process Theory for Multitimescale Embodied Collective Cognitive Systems.

Artificial life·2025
Same journal

To Engineer an Angel, First Validate the Devil: Analyzing the "Could Be" in Artificial Life's "Life as-It-Could-Be".

Artificial life·2025
Same journal

Untapped Potential in Self-Optimization of Hopfield Networks: The Creativity of Unsupervised Learning.

Artificial life·2025
See all related articles

Generative representations improve evolutionary robotics by enabling element reuse, leading to faster evolution of more capable robots. This approach captures design biases and allows larger mutations for enhanced performance.

Area of Science:

  • Evolutionary Computation
  • Robotics
  • Artificial Intelligence

Background:

  • Scalability in body-brain evolution systems is limited by creature encoding methods.
  • Current representations often lack efficient element reuse between genotype and phenotype.

Purpose of the Study:

  • To define and demonstrate generative representations for evolutionary systems.
  • To introduce GENRE, an evolutionary system utilizing generative representations.
  • To compare generative representations against direct representations in robotics evolution.

Main Methods:

  • Defined generative representations based on genotype-phenotype element reuse.
  • Developed GENRE, an evolutionary system for concurrent morphology and neural controller evolution.
  • Evolved simulated robots for locomotion using GENRE and a direct representation.

Related Experiment Videos

Main Results:

  • The generative representation system (GENRE) evolved robots with significantly higher fitness.
  • GENRE achieved superior performance by leveraging design space biases.
  • Viable large-scale mutations in the phenotype were enabled by the generative approach.

Conclusions:

  • Generative representations enhance evolutionary robotics by facilitating encapsulation, coordination, and reuse of parts.
  • This representation class offers a scalable solution for complex evolutionary systems.
  • Generative representations lead to more efficient and effective evolutionary design.