Creating High-Level Components with a Generative Representation for Body-Brain Evolution

@article{Hornby2002CreatingHC,
  title={Creating High-Level Components with a Generative Representation for Body-Brain Evolution},
  author={Gregory Hornby and Jordan B. Pollack},
  journal={Artificial Life},
  year={2002},
  volume={8},
  pages={223-246}
}
One of the main limitations of scalability in body-brain evolution systems is the representation chosen for encoding creatures. This paper defines a class of representations called generative representations, which are identified by their ability to reuse elements of the genotype in the translation to the phenotype. This paper presents an example of a generative representation for the concurrent evolution of the morphology and neural controller of simulated robots, and also introduces GENRE, an… 

Generative representations for the automated design of modular physical robots

This work demonstrates an automatic design system that produces complex robots by exploiting the principles of regularity, modularity, hierarchy, and reuse, and demonstrates for the first time the evolution and construction of modular, three-dimensional, physically locomoting robots.

Generative Representations for Evolving Families of Designs

An evolutionary design system is described which uses a generative representation to encode families of designs and is demonstrated on two design substrates: neural-networks which solve the 3/5/7-parity problem and three-dimensional tables of varying heights.

Creating Complex Building Blocks through Generative Representations

This work defines generative representations as those which have the ability to reuse elements in an encoding through either iteration or abstraction and argues that reuse improves functional scalability by allowing the representation to construct buildingblocks and capture design dependencies.

Evolving artificial neural networks with generative encodings inspired by developmental biology

The general conclusion that can be drawn from this work is that generative encodings can produce some of the properties seen in complex, natural organisms, and will likely be an important part of the long-term goal of synthetically evolving phenotypes that approach the capability, intelligence, and complexity of their natural rivals.

Topological reasoning using a generative representation and a genetic algorithm

This thesis improves the understanding of generative representations and contributes to the knowledge that is required to further develop them for real-world use.

Comparing the Evolvability of Generative Encoding Schemes

A novel approach to measure the evolvability provided by an encoding is reported, by characterizing both the quality of the mutations and the quantity of phenotypic variation, and the number of generators required by amputated individuals to recover an effective gait is indicated.

An evolutionary model for 3 D agents integrating continuous and plastic development

A grammar based approach for generating agents with both 3D morphology and neural networks that leads to more natural developmental processes than previous approaches is introduced.

The Development of Modular Evolutionary Networks for Quadrupedal Locomotion

This work allows the neural network to grow, building itself up, piece by piece, from a simple to a complex form, by allowing the robot’s body plan and environment to develop while simultaneously adding to the structure of the controlling network.

Functional Scalability through Generative Representations: The Evolution of Table Designs

It is argued that generative representations, those which are capable of reusing elements of the encoded design in the translation to the actual artifact, are better suited for automated design because reuse of building blocks captures some design dependencies and improves the ability to make large changes in design space.

A unified approach to evolving plasticity and neural geometry

The most interesting aspect of this investigation is that the emergent neural structures are beginning to acquire more natural properties, which means that neuroevolution can begin to pose new problems and answer deeper questions about how brains evolved that are ultimately relevant to the field of AI as a whole.
...

References

SHOWING 1-10 OF 54 REFERENCES

Generative representations for the automated design of modular physical robots

This work demonstrates an automatic design system that produces complex robots by exploiting the principles of regularity, modularity, hierarchy, and reuse, and demonstrates for the first time the evolution and construction of modular, three-dimensional, physically locomoting robots.

Evolution of generative design systems for modular physical robots

  • G. HornbyHod LipsonJ. Pollack
  • Computer Science
    Proceedings 2001 ICRA. IEEE International Conference on Robotics and Automation (Cat. No.01CH37164)
  • 2001
It is claimed that for automatic design systems to scale in complexity the designs they produce must be made of re-used modules, based on a generative design grammar subject to an evolutionary process.

The advantages of generative grammatical encodings for physical design

  • G. HornbyJ. Pollack
  • Computer Science
    Proceedings of the 2001 Congress on Evolutionary Computation (IEEE Cat. No.01TH8546)
  • 2001
This work describes a system for creating generative specifications by combining Lindenmayer systems with evolutionary algorithms, and applies it to the problem of generating table designs and finds that the generative system produces designs with higher fitness and is faster than the non-generative system.

Automated learning of muscle-actuated locomotion through control abstraction

We present a learning technique that automatically syn- thesizes realistic locomotion for the animation of physics-based models of animals. The method is especially suitable for animals with highly

Evolving virtual creatures

A genetic language is presented that uses nodes and connections as its primitive elements to represent directed graphs, which are used to describe both the morphology and the neural circuitry of creatures that move and behave in simulated three-dimensional physical worlds.

Functional Scalability through Generative Representations: The Evolution of Table Designs

It is argued that generative representations, those which are capable of reusing elements of the encoded design in the translation to the actual artifact, are better suited for automated design because reuse of building blocks captures some design dependencies and improves the ability to make large changes in design space.

Evolving 3D Morphology and Behavior by Competition

This article describes a system for the evolution and coevolution of virtual creatures that compete in physically simulated three-dimensional worlds that can adapt to each other as they evolve simultaneously.

Comparison of Different Genotype Encodings for Simulated Three-Dimensional Agents

The effect of different genetic encodings used for evolving three-dimensional agents with physical morphologies, including a low-level direct mapping and two higher-levelencodings, are analyzed.

Evolving L-systems to generate virtual creatures

The World of Framsticks: Simulation, Evolution, Interaction

A three-dimensional virtual world simulation is described, where evolution takes place and it is possible to investigate behaviors of creatures in real-time and mimic spontaneous evolution when the fitness is defined as the life span of the organisms.
...