Research on
Virtual Creatures ( Virtual Life)
A Proposal
by Ivan Tanev & Tom Ray
Vision
The
products of natural adaptive systems (such as living nature and the human mind)
vastly exceed the complexity of anything created by humans. Only when we identify and harness the
principles behind natural adaptive systems will we be able to create human
artifacts of similar adaptive complexity.
Several decades of exponential growth in computing power has brought us
to the threshold of a new era in computation and has led to the ubiquitous
presence of computing machines containing vast numbers of components. The luxury of this rich bounty of computing
elements invites the creation of new paradigms in computation. The traditional and still ubiquitous Von
Neumann computer is rigid and brittle.
By its very design, it lacks the robust and adaptive quality of living
organisms evolved and grown from genomes.
Cross-fertilization between biology and computer science has opened the
era of evolvable software and hardware, while at the same time the development
of re-configurable chips is blurring the distinction between hardware and
software. In a world where both genomic
information and digital computing elements have reached the explosive phase of
their exponential growth, where life can be seen as a form of computation and
some forms of computation can be seen as life, where evolution can inhabit both
the organic and the digital medium, an era of active exploration of
fundamentally new, and often biologically inspired computer architectures has
begun.
Objectives
·
Better
performing, aesthetically and/or functionally
·
Faster and more
richly evolving
·
More robust and
adaptive
·
More efficiently
implemented in hardware
Research Directions
The
ability of VC&R system to evolve both morphology and behavior, situated in
realistically simulated environments, offers the opportunity to conduct
research in many directions, including the following:
·
Symmetry the
evolution/emergence of planes of morphological symmetry in natural species (fighting
gravity? locomotive efficiency? genome compression? optimum performance? checksum
for tolerance to partial genetic damage? ...)
·
Epigenesis development of VC&R through variable gene
expression mechanisms
·
Allometry growth of a part of an organism in relation to the growth
of the whole organism or some part of it
·
Heterochrony a genetic shift in timing of the development of a tissue
or anatomical part, or in the onset of a physiological process, relative to an
ancestor
·
Parsimony principle
applied to sensory abilities of VC&R (e.g. smell vs. color vision)
·
Differentiation
how functional specialization in multi-agent VC&R builds morphological
specialization
·
Polymorphism
the comparative study of the presence of two or more distinct phenotypes in a
population due to the expression of different alleles of a given gene
·
Polyphenism polymorphism due to environmental changes
·
Epigenesis polymorphism and polyphenism
through epigenesis using controllable histone code and/or genetic regulatory
networks
·
Communication
the emergence and survival value of morphological traits, relevant to signaling
among VC&Rs
·
Coevolution study of evolutionary interactions between multiple
populations sharing an environment
·
Social behavior
study of cooperative, competitive, amorous and other interactions between
individuals of the same species
·
Directed graph VC&R
systems utilize a flexible and powerful directed-graph representation of both
morphology and neural circuitry.
·
Evolvable The
neural circuitry of VC&R is completely distributed, arbitrarily complex,
yet seemingly highly evolvable.
·
Universal
controller architecture The directed graph of the neural circuitry defines
the connectivity of the sensory inputs, the output actuators, and the
processing elements, and may be considered to be a universal paradigm for
building robot controllers.
·
Hardware evolution
Such a nervous system might have a straightforward mapping into hardware,
suggesting a possible implementation of artificial brain as an evolvable
hardware.
Required Resources