Evolving 3D objects with a generative encoding inspired by developmental biology

@article{Clune2011Evolving3O,
  title={Evolving 3D objects with a generative encoding inspired by developmental biology},
  author={Jeff Clune and Hod Lipson},
  journal={ACM Sigevolution},
  year={2011},
  volume={5},
  pages={2-12}
}
This paper introduces an algorithm for evolving 3D objects with a generative encoding that abstracts how biological morphologies are produced. Evolving interesting 3D objects is useful in many disciplines, including artistic design (e.g. sculpture), engineering (e.g. robotics, architecture, or product design), and biology (e.g. for investigating morphological evolution). A critical element in evolving 3D objects is the representation, which strongly influences the types of objects produced. In… 

A Compiler for CPPNs: Transforming Phenotypic Descriptions Into Genotypic Representations

  • S. Risi
  • Computer Science
    AAAI Fall Symposia
  • 2013
The concept of a CPPN-Compiler, which allows the user to directly compile a high-level description of the desired starting structure into the CPPNs itself, is introduced, which opens up a new research direction in GDS, in which specialized CPPn-Compilers for different domains could help to overcome the black box of evolutionary optimization.

Convergent evolution in silico reveals shape and dynamic principles of directed locomotion

It is found that an intermediate number of body modules (appendages) and high body symmetry are evolutionarily selected regardless of gravitational environments, robot sizes, and genotype encoding.

Evolving Morphologies with CPPN-NEAT and a Dynamic Substrate

An extension of CPPN-NEAT is presented, in which nodes grow connections across a dynamic substrate, and this model outperforms similar HyperNEAT methods for approximating specific connectivity patterns, and suggests important clues regarding how to best harness generative and developmental representations to build scalable and high-performance physical morphologies.

Automatically Designing and Printing 3-D Objects with EvoFab 0.2

EvoFab 0.2 is described, a completely automated physically embodied machine which implements Evolutionary Fabrication and evolves three dimensional objects and is shown how it can be used to create novel structures.

The Effects of Learning in Morphologically Evolving Robot Systems

Using extensive simulations, it is shown that learning can greatly increase task performance and reduce the number of generations required to reach a certain fitness level compared to the purely evolutionary approach, and that the evolved morphologies will be also different, even though learning only directly affects the controllers.

Evolutionary Fabrication: An Autonomous System of Invention

EvoFab is a machine built upon a process that can, in principle, automatically invent and build anything, from soft robots to new toys, by evolving the process, not the product.

1D Printing of Recyclable Robots

A 1-D printing system that uses an approach inspired by the ribosome to fabricate a variety of specialized robotic automata from a single string of source material, enabling an autonomous manufacturing ecosystem capable of repurposing previous iterations to accomplish new tasks.

Graph Grammars as a Representation for Interactive Evolutionary 3D Design

A new interactive evolutionary 3D design system based on graph grammars, a fascinating and powerful formalism in which sub-graphs, nodes and edges are iteratively rewritten by rules analogous to those of context-free Grammars and shape grammar, demonstrates the flexibility of the representation.

One-Dimensional Printing of Recyclable Robots

A 1-D printing system that uses an approach inspired by the ribosome to fabricate a variety of specialized robotic automata from a single string of source material, enabling an autonomous manufacturing ecosystem capable of repurposing previous iterations to accomplish new tasks.

Shape Optimization With Surface-Mapped CPPNs

This paper presents a new evolutionary approach to shape optimization using what it calls “surface-mapped compositional pattern producing networks (CPPNs),” which outperforms a state-of-the-art gradient-based method on a simple benchmark problem, and scales well as degrees of freedom are added to the design problem.
...

References

SHOWING 1-10 OF 66 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.

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.

A Taxonomy for Artificial Embryogeny

This taxonomy provides a unified context for long-term research in AE, so that implementation decisions can be compared and contrasted along known dimensions in the design space of embryogenic systems, and allows predicting how the settings of various AE parameters affect the capacity to efficiently evolve complex phenotypes.

Morphological evolution of freeform robots

The evolution of locomoting amorphous robots composed of multiple materials is demonstrated and the results open the door to a new design space that more closely mimics the freeform,Amorphous and continuous nature of biological systems.

Dynamic Resolution in the Co-Evolution of Morphology and Control

This paper presents a novel method for co-evolving morphology and control using CPPN-NEAT, capable of dynamically adjusting the resolution at which components of the robot are created.

Evolving CPPNs to grow three-dimensional physical structures

A novel method for evolving three-dimensional physical structures using CPPN-NEAT is introduced which is capable of producing artifacts that capture the non-obvious yet close relationship between function and physical structure.

How crystals that sense and respond to their environments could evolve

An enduring mystery in biology is how a physical entity simple enough to have arisen spontaneously could have evolved into the complex life seen on Earth today. Cairns-Smith has proposed that life

Repeated structure and dissociation of genotypic and phenotypic complexity in artificial ontogeny

It is demonstrated that evolved genetic regulatory networks in AO give rise to hierarchical, repeated phenotypic structures, and the claim that artificial ontogeny is a useful design tool for the evolutionary design of virtual agents and real-world robots is supported.

The sensitivity of HyperNEAT to different geometric representations of a problem

The results suggest that HyperNEAT practitioners can obtain good results even if they do not know how to geometrically represent a problem, and that further improvements are possible with a well-chosen geometric representation.

An information-bearing seed for nucleating algorithmic self-assembly

This work demonstrates how DNA origami seeds enable the easy, high-yield, low-error-rate growth of algorithmic crystals as a route toward programmable bottom-up fabrication.
...