in Artificial Life VIII, Standish, Abbass, Bedau (eds)(MIT Press) 2002. pp 101–105
1
Embryonics: Electronic Stem Cells
Lucian Prodan1 , Gianluca Tempesti2 , Daniel Mange2 , and André Stauffer2
1
2
“Politehnica” University (UPT), Timisoara, Romania
E-mail: [email protected], WWW: http://www.cs.utt.ro
Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland
E-mail: [email protected], WWW: http://lslwww.epfl.ch
Abstract
Embryonics is a long-term research project attempting
to draw inspiration from the biological process of ontogeny, to implement novel digital computing machines
endowed with better fault-tolerant capabilities. This
article discusses the degree of bio-inspiration attained
while also attempting to start a similarity debate on
various implementation decisions, on why and how nature developed its subtle, intricated means of growing,
healing and reproducing.
Motto: “What makes stem cells special is that
they’re immortal, and they can become anything
they want to be.” — Dr. James Thomson, University of Wisconsin.
Introduction
The incredibly huge number of some 60 trillion (60 ×
1012 ) cells make up a human being, with as many as
10 billion (1010 ) cells with 100 trillion (1014 ) interconnections concentrated in each of our brains (Mange &
Tomassini 1998). Intelligence, creativity, the capacity of
abstractization and ultimately conscience are all possible thanks to this marvelously complex human machine.
Yet its entire structure emerges from a single cell, the
zygote, giving birth to a completely functional organism
that will continue to develop and enhance its features
throught its entire life.
However, there is a key question that arises, driving
biologists and not only: how can all this be possible?
How can this be, that a single cell would divide for such
a large number of times even us humans find difficult to
imagine? And what mechanisms direct the division process so perfectly that when it ends the result is a healthy
organism? Would it be possible to exploit these mechanisms to inspire engineers, perhaps including novel computational implementations? It seems that nature has
created a remarkable circle: we became complex enough
to develop sciences and now scientists have reached the
point that enables them to study how nature can reach
such complexity.
A special type of cell appears to answer some of the
previous questions, a cell that can give birth to other
identical cells, and all being able to become specialized
cells themselves, such as muscular or nerve cells. But
it is not all that easy to discover nature, let alone to
penetrate its mysteries. After nearly 20 years of hard
work, two teams of scientists have succeeded in growing
and replicating these mother cells (Thomson & others
1998). Called “stem cells”, these basic units ultimately
mature and differentiate to become the building material
of all types of body tissue (May 2000).
The fact that scientists are all driven by common questions is remarkably demonstrated by recent history, the
research on novel bio-inspired computing systems having roughly the same age as modern biology. In the late
1940’s von Neumann began to develop a theory of automata in order to contribute to a better understanding
of both natural automata (living beings) and of artificial automata (computational systems) (Sipper et al.
1998). In more recent years there has been significant
research carried out in a new project, aiming at creating bio-inspired computer hardware, called Embryonics
(Tempesti 1998; Tempesti, Mange, & Stauffer 1998).
The purpose of this article is not to delve into the
ethical issues’ debate — related to cloning or using the
DNA molecule as a threat (Alexander 2000) — but to
take a comparative look to the recent findings in the field
of cellular biology and their virtual mirror, the world of
Embryonics. Intriguingly, there are many more similarities and common ideas than it is possible to explain by
a mere coincidence. Section 2 briefly describes the discovery of a new kind of cell. Section 3 presents the path
to a new kind of bio-inspired computer hardware within
the Embryonics research environment, together with a
deeper look inside the characteristics of the newly discovered cell and how the two relate to each other. Finally, Section 4 presents the conclusions and some general guidelines for the future of the Embryonics.
Nature Reveals Its Ways
Each and every living multi-cellular organism consists of
groups of specialized cells that make up tissues and organs. Throughout its entire life, the perfect operation of
this magnificent massively parallel biological system is
transforming capacity could provide a solution for an
engineering.
organism’s survival to extreme stress such as loss of a
vital organ, giving it a huge potential in biology and
medicine. Because this cell is the precursor to all other
3. The Embryonics Project − Biological
cell
has set
2 types in the human body, thisinaccomplishment
Artificial Life VIII,
Standish, Abbass, Bedau (eds)
(MIT Press) 2002. pp 101–105
Connections
the stage for a revolution in medicine and basic biology
[16, 18].
As a long term research project, Embryonics has been
developing
some
years now
becoming
a consistent
specific
type offor
cell.
Research
conducted
at Stanford
rerepository
of
bio−inspired
solutions
in
computer
veals that in fact there are some stem cell “guardians”,
engineering. Along with the proposed POE model [12],
surrounding areas composed of stem cells, that deterthe Embryonics project aims at developing ontogenetic
mine which type of ordinary cell they will specialize in
hardware, i.e. hardware capable of self−replicating,
(Vaughan
2000). and evolving from its very own genetic
self−repairing
As
far
as
material.we know, the most interesting characteristics
of the
stem
cells arehas
thebecome
following:
Fault
tolerance
a major requirement in
modern
computing.
A
great
deal
of research effort has
• they can give rise to specialized cells;
been spent to achieve this feature and considerable
boost comes they
from seem
technological
• undifferentiated,
to have theadvancements,
ability to dienabling
system
designers
to inconsider
vide for apparently indefinite periods
culture; new
reconfiguration techniques applied to an entire
computing
universe
made bydevelop
connecting
many
• any
of these cells
can potentially
into a fetus.
identical entities [4, 5, 7]. While the struggle for
Fig. 1. The 4 levels of organization of Embryonics.
These
are sufficient
arguments
the fantastic
performance
is still
present, to
theconsider
term’s definition
has
Figure 1: The 4 levels of organization of Embryonics.
potential
stem cellsloose
that due
lies toin atheir
abilitylevel
to be
becomeofsomewhat
sufficient
of
The zygote has the potential to form an entire organism
ultimately
directed force
to become
specific
types any
of cells
or
brute computing
available
from nearly
digital
as it has the remarkable feature of being totipotent,
tissue.
If, onToday
one hand,
this feature
could
be the
usedfirst
to
machine.
we witness
the shift
toward
meaning
that
its potential
is of
total
[22]. division,
Immediately
ensured by
a ceaseless
process
cellular
actitreat
a hostofofthe
cell-based
on the
other
hand it
priority
quest fordiseases,
distributed,
highly
redundant
following
the fertilization, it starts to divide into identical
vated in order to overcome wounds and illnesses. A great
andhave
massively
reconfigurable
and the
might
a significant
impact on the systems
world of computer
totipotent cells that begin to specialize after several cycles
majority of such cases are successfully repaired, with the
Embryonics project makes no exception. Recent
engineering.
of cell division, forming a hollow sphere of cells, called a
reliability analysis also show that embryonic structures
exception
of extreme situations such as the complete deblastocyst [19].
open
a new direction for
building —
highly
capable fault
The
Embryonics
Project
Biological
struction
of
a
vital
organ
or
tissue.
When
this
occurs,
But mastering the conditions necessary to witness the
tolerant
systems
[6].
there
are
no
possibilities
of
healing,
which
leads
to
the
Connections
delicate phenomenon of cellular division is no simple
Characteristics such as replication (self−replication),
death ofYet
the entire
organism. made at Johns Hopkins
matter.
the experiments
As awhich
long term
research
project,
Embryonics
been and
decan be seen as
a special
case of has
growth,
University
successfully
verified
that stem originates
cells have from
the
At its very
beginning,
each organism
veloping
for
some
years
now
becoming
a
consistent
reposregeneration (self−repair), or recovery after wounds or
true
potential
of developing
basic cells process
found intakes
all
a single
cell, the
zygote. Itsinto
development
itory
of bio-inspired
in computer
engineering.
illnesses,
are partsolutions
of the ontogeny
process
and are
mammalian
embryos
[17].cellular
However,
the
place through
successive
divisionsdirecting
and differentiAlong
with
the
proposed
POE
model
(Sipper
et al. 1997,
extremely
attractive
for
many
applications.
The
specialization
process
of the ofstem
is cell,
another
ations, proof for
the existence
a verycells
special
capaalsoEmbryonics
this volume,project
p??), the
Embryonics
project view
aims at
presents
a consistent
of
problem.
We
do
know
now
that
stem
cells
can
in
fact
ble of differentiating into any kind of needed, specialized
developing
ontogenetic
ontogenetic
hardwarehardware,
and beyondi.e.
[2,hardware
11, 14]. capable
become any type of cell, either resulting new stem cells or
cell. Christened stem cell, its unique transforming capacEmbryonics quasi−biological
organisms
of self-replicating,
self-repairing artificial
and evolving
from are
its
specialized cells. What we are uncertain of is what
ity
could
provide
a
solution
for
an
organism’s
survival
to
made
up
of
a
finite
number
of
functionally
identical
cells
very
own
genetic
material.
exactly instructs a stem cell to specialize into a specific
extreme
stress
such asconducted
loss of a at
vital
organ,reveals
givingthat
it a
[3,Fault
4, 13,tolerance
14], eachhas
of which
is ina turn
made
up of a finite
become
major
requirement
in
type
of cell.
Research
Stanford
huge
biology
and
medicine.
Because
this cell
number computing.
of functionally
identical
1).
in
factpotential
there areinsome
stem
cell
"guardians",
surrounding
modern
A great
deal molecules
of research(Figure
effort has
Eachspent
cell toisachieve
a simple
processor
binary decision
is the composed
precursor to
other
cellthat
types
in the human
areas
of all
stem
cells,
determine
which body,
type
been
this feature
and(a
considerable
boost
machine)
realizing
a
unique
function
within
the organism,
this
accomplishment
hasspecialize
set the stage
for a revolution in
of
ordinary
cell they will
in [20].
comes from technological advancements, enabling
sysdefined by a set of instructions (program), which we call
As
far as we
most interesting
of
medicine
andknow,
basicthe
biology
(Thomsoncharacteristics
& others 1998;
tem designers to consider new reconfiguration techniques
the gene of the cell. The functionality of the organism is
Anonymous 2001).
applied to an entire computing universe made by conThe zygote has the potential to form an entire organnecting many identical entities (Mange et al. 1998;
ism as it has the remarkable feature of being totipotent,
Negrini, Sami, & Stefanelli 1989; Shibayama et al. 1997).
meaning that its potential is total (Anonymous 2000).
While the struggle for performance is still present, the
Immediately following the fertilization, it starts to diterm’s definition has become somewhat loose due to a
vide into identical totipotent cells that begin to specialsufficient level of brute computing force available from
ize after several cycles of cell division, forming a hollow
nearly any digital machine. Today we witness the shift
sphere of cells, called a blastocyst (May 2000).
toward the first priority of the quest for distributed,
highly redundant and massively reconfigurable systems
But mastering the conditions necessary to witness the
and the Embryonics project makes no exception. Recent
delicate phenomenon of cellular division is no simple
reliability analysis also show that embryonic structures
matter. Yet the experiments made at Johns Hopkins
open a new direction for building highly capable fault
University successfully verified that stem cells have the
tolerant systems (Ortega & Tyrrell 1999).
true potential of developing into basic cells found in all
Characteristics such as replication (self-replication),
mammalian embryos (Shamblott & others 1998). Howwhich can be seen as a special case of growth, and reever, directing the specialization process of the stem cells
generation (self-repair), or recovery after wounds or illis another problem. We do know now that stem cells
nesses, are part of the ontogeny process and are excan in fact become any type of cell, either resulting new
tremely attractive for many applications. The Emstem cells or specialized cells. What we are uncertain of
bryonics project presents a consistent view of ontogeis what exactly instructs a stem cell to specialize into a
a rectangular cellular membrane, specified by a special
part of the genome, called the polymerase genome. Each
cell determines which gene is to be executed based on its
coordinates inside the organism and stores a complete
copy of the operative genome, thus making possible any
in Artificial
Standish,
(eds)(MIT
task
transfer. Life
ThisVIII,
feature
makes Abbass,
the cellsBedau
capable
of
virtually replacing each other (in the event of a detected
malfunction) just like in real biological organisms.
The hardware
process ofand
a cell
deciding
which
geneWolpert
to execute
netic
beyond
(Gilbert
1991;
1991;
based
the 2001).
coordinates from within its local
Prodanonet al.
environment, i.e. the surrounding cells, determines its
Embryonics quasi-biological artificial organisms are
functionality, not unlike cellular specialization in
made
up organisms.
of a finite Furthermore,
number of functionally
biological
the access identical
to the
cells
(Mange
&
Tomassini
1998;
Mange
et cell
al. with
1998;
whole genetic program provides our electronic
Prodan et al.much
2000;as2001),
each ofstem
whichcells
is in have
turn made
universality,
biological
the
up of a finite
number ofany
functionally
identical molecules
potential
of becoming
type of specialized
cell.
(Figure
1). no
Each
cellon
is ahow
simple
(a binary
deciThere are
limits
largeprocessor
Embryonics
cells can
sion
machine)
realizing of
a unique
be.
Then,
the universality
the cellfunction
becomeswithin
actuallythe
theorcapacity
of executing
variable
tasks (program),
− the more
ganism, defined
by a set of
instructions
which
computationally
complex
the
task,
the
more
molecules
we call the gene of the cell. The functionality
of the orrequired
the cell structure.
Furthermore,
carefully of
ganism for
is therefore
obtained by
the parallelbyoperation
selecting
which
portion
of
the
code
is
to
be
executed
all the cells while all the genes make up the genetic by
promolecules,
their
universality
is also
assured
− a molecule
gram or the
operative
genome.
Cells
are delimited
by the
can effectively replace any other one by simply adapting
existence of a rectangular cellular membrane, specified
its internal code.
by
a special
of the
genome,
the polymerase
Every
livingpart
cell’s
innermass
is called
delimited
from the
genome.
Each
cell
determines
which
gene
to be exesurrounding environment by the cellular is
membrane,
cuted
based
on
its
coordinates
inside
the
organism
and
which also acts as an interface with the exterior, allowing
stores
a
complete
copy
of
the
operative
genome,
thus
a limited exchange of substances. If the biological world
making
transfer.
This feature
makes
allows
andpossible
dependsany
on task
exchanging
substances,
the world
cells has
capable
virtuallyrules:
replacing
each other
(in the
ofthe
silicon
more of
restrictive
the material
replacing
substances,
nonetheless
allowedjust
and like
depended
its
event of a but
detected
malfunction)
in realon
biologexchanging
is information. Much as in nature, where
ical organisms.
substances
entertain
lifedeciding
by carrying
energy
and
The process
of a cell
which gene
to execute
information, in the world of silicon electronic signals
based on the coordinates from within its local environcarry in a similar way the same ingredients, entertaining
ment,
artificiali.e.
life. the surrounding cells, determines its functionality,
not unlike
cellular
in divider)
biological
The artificial
membrane
(alsospecialization
called the space
organisms.
Furthermore,
the
access
to
the
whole
genetic
has a triple role (Figure 2). Firstly, it acts like a spatial
programlogically
provides our
electronic
cell with (molecules)
universality,
barrier,
separating
resources
much as biological
the potential
of bebelonging
to differentstem
cellscells
andhave
ensuring
an individual
identity
thetype
surrounding
environment.
cominginany
of specialized
cell. Secondly, it acts
as aThere
guide are
for no
thelimits
entering
configuration,
which contains
on how
large Embryonics
cells can
the
partuniversality
of the genome
(genes
A to actually
D for
be.operative
Then, the
of the
cell from
becomes
the
left
cell
and
from
G
to
L
for
the
right
cell,
Figure
2).
the capacity of executing variable tasks — the more comAll molecules pertaining to a cell are configured with the
putationally complex the task, the more molecules recorresponding gene in a chain−like process, which does
quired
forinformation
the cell structure.
Furthermore,
by and
carefully
not allow
to get outside
the cell
be
selecting
which
portion
of
the
code
is
to
be
executed
wasted; in a similar manner, the existence of the cellularby
molecules,intheir
universality
also assured
— a molecule
membrane
biological
cells is
restricts
the access
of the
can
effectively
replace
any
other
one
by
simply
adapting
environment to their inner part both ways, thus preventing
its unwanted
internal code.
any
loss of substances or possible intrusions to
or from
the
environment.
Thirdly,is its
presencefrom
triggers
a
Every living cell’s innermass
delimited
the surmechanism
determining
which
gene
is
to
be
executed
by
rounding environment by the cellular membrane, which
each
can bewith
seenthe
asexterior,
a specialization
alsomolecule.
acts as anThis
interface
allowingatathe
limited exchange of substances. If the biological world allows and depends on exchanging substances, the world of
silicon has more restrictive rules: the material replacing
substances, but nonetheless allowed and dependent on
its exchange, is information. Much as in nature, where
substances entertain life by carrying energy and information, in the world of silicon electronic signals carry in a
similar way the same ingredients, entertaining artificial
life.
The artificial membrane (also called the space divider)
has a triple role (Figure 2). Firstly, it acts like a spatial barrier, logically separating resources (molecules)
What makes living beings so complex is the subtle
cooperation between various mechanisms helping to
preserve the innate features of an organism and
continuously adapting them to a quite considerable extent
to environmental challenges. Life is a quite dramatic and
Press)
2002. process,
pp 101–105
long testing
so it is perhaps best for biological3
entities to employ self−developed procedures in order to
check for abnormalities and trigger the repair processes.
Fig. 2.2:Self−repair
thethe
molecular
level.
Figure
Self-repairat at
molecular
level
As in nature, where multiple self−testing happens in each
belonging
to different
and ensuring
an individual
and
every living
being, cells
Embryonics
also relies
on more
identity
in the
surrounding
than
one such
mechanisms
[4, environment.
9, 10]. The verySecondly,
first self−it
acts as procedure
a guide foremploys
the entering
configuration,
which
testing
test−vectors
and applies
to conthe
core
theoperative
molecule.part
Anofoff−line
testing(genes
procedure
tainsofthe
the genome
from was
A to
considered
to becellsufficient,
place
D for the left
and fromand
G takes
to L for
thebefore
right any
cell,
critical
data
is
loaded.
Nature
does
a
similar
process
at the
Figure 2). All molecules pertaining to a cell are configlowest
level
contained
very intimate
ured with
thepossible,
corresponding
gene by
in athe
chain-like
process,
structure
of
the
DNA
itself.
The
two
strands
that make
which does not allow information to get outside
the up
cell
the
continuously
eachmanner,
other by the
subtle
chemicalof
andDNA
be wasted;
in a test
similar
existence
bonds,
preventing
and detecting
a majority
of possible
the cellular
membrane
in biological
cells restricts
the acerrors.
During
operation,
the
molecular
core
can
act as
an
cess of the environment to their inner part both
ways,
active memory, much as the DNA does. The self−testing
thus preventing any unwanted loss of substances or posis ensured by "breaking" its internal register into two
sible intrusions to or from the environment. Thirdly, its
halves storing complementary data and acting in a similar
presence
triggers
a mechanism
way
the two
DNA strands
do [14].determining which gene
is Another
to be executed
by
each
molecule.
can beunit,
seen isas
part of the molecule,
the This
functional
a
specialization
at
the
molecular
level,
the
surrounding
implemented as an on−line self−testing device, using the
membrane
directing
the whole
in a similar
voting
majority
technique.
Onceprocess
the testing
features way
in
the
stem
cells’
“guardians”
(Vaughan
2000)
control cell
place, any detected error should allow the recovery,
or
specialization
biology. Due to the vast complexity of
healing
of the in
organism.
What makes
living beingshealing
so complex
is the subtle
biological
organisms,
(self−repairing)
mechanisms
can only
be mechanisms
effective if helping
the task
is
cooperation between
various
to prehierarchically
decomposed
dealt
with accordingly.
serve the innate
features and
of an
organism
and continuEmbryonics
a hierarchy
of two
self−to
ously
adaptinguses
them
to a quitecomposed
considerable
extent
repairing
mechanisms.
The Life
cellular
is built and
of
environmental
challenges.
is astructure
quite dramatic
active
and spare
molecules.
the event
long testing
process,
so it is In
perhaps
best of
forfunctional
biological
failure
active molecule,
all its procedures
functions areintaken
entitiesoftoanemploy
self-developed
orderbyto
the
nearest
spare
molecule
by
means
of
shifting
part
of the
check for abnormalities and trigger the repair processes.
cellular
active
resources
one
position.
This
is
the
self−in
As in nature, where multiple self-testing happens
repairing mechanism at the molecular level (Figure 2). A
each and every living being, Embryonics also relies on
similar process takes place in nature, where the cell is
more than one such mechanisms (Mange et al. 1998;
capable of fabricating the resources needed. This is
Tempesti 1998;
Tempesti,
Stauffer 1998).
The
obviously
impossible
with Mange,
current &
technology,
the only
very
first
self-testing
procedure
employs
test-vectors
and
way of providing additional resources at the molecular
applies to the core of the molecule. An off-line testing procedure was considered to be sufficient, and takes
place before any critical data is loaded. Nature does a
similar process at the lowest level possible, contained by
the very intimate structure of the DNA itself. The two
strands that make up the DNA continuously test each
other by subtle chemical bonds, preventing and detecting a majority of possible errors. During operation, the
molecular core can act as an active memory, much as the
DNA does. The self-testing is ensured by “breaking” its
internal register into two halves storing complementary
data and acting in a similar way the two DNA strands
4
in Artificial Life VIII, Standish, Abbass, Bedau (eds) (MIT Press) 2002. pp 101–105
do (Prodan et al. 2001).
Another part of the molecule, the functional unit, is
implemented as an on-line self-testing device, using the
voting majority technique. Once the testing features in
place, any detected error should allow the recovery, or
healing of the organism. Due to the vast complexity
of biological organisms, healing (self-repairing) mechanisms can only be effective if the task is hierarchically
decomposed and dealt with accordingly.
Embryonics uses a hierarchy composed of two selfrepairing mechanisms. The cellular structure is built of
active and spare molecules. In the event of functional
failure of an active molecule, all its functions are taken
by the nearest spare molecule by means of shifting part
of the cellular active resources one position. This is the
self-repairing mechanism at the molecular level (Figure
2). A similar process takes place in nature, where the cell
is capable of fabricating the resources needed. This is obviously impossible with current technology, the only way
of providing additional resources at the molecular level
being as spares. During its artificial life, the electronic
organism may experience the situation when inside a cell
a new fault is detected, but no more spares are available. This is the moment the second self-repair mechanism enters the stage, at the cellular level: the faulty
cell “dies”, then it is isolated, and all active cells are
shifted a position by rerouting all the interconnections
(Mange et al. 1998; Negrini, Sami, & Stefanelli 1989).
This is again inspired by nature, where foreign bodies
(objects or mutating cells) are isolated and eventually
eliminated. Self-replication can be seen as a special
case of growth; after configuring an initial cellular structure, the self-replication process considers this as a pattern and colonizes the whole environment (Sipper 1998;
Tempesti 1998).
Conclusions
This article presented the essential bio-inspired features
of the Embryonics project. It all started as a long-term
research project, and, while it still remains largely so,
the bio-inspired hardware developed allows us to build
computing machines endowed with improved robustness,
copied and adapted from biology to electronics. The
technological advancements also give us now the opportunity to implement digital systems based on novel, complex operating principles but also put a higher pressure
onto the design process and necessary trade-offs. Successful bio-inpired hardware may emerge only together
with successful theory on nature’s ways and the ways of
implementing them in silicon.
All discussed issues lead us to believe that real bioinspired hardware systems might just be closer than we
think, thus narrowing the gap between biology and biologically inspired digital systems.
References
Alexander, D. 2000. The human genome project. Science & Christian Belief 12(2):98.
Anonymous.
2000.
Stem cells:
A primer.
Technical report, National Institutes of Health.
http://www.nih.gov/news/stemcell/primer.htm.
Anonymous. 2001. Opportunities and challenges: A focus on future stem cell applications. In Stem Cells:
Scientific Progress and Future Research Directions.
National Institutes of Health, Department of Health
and Human Services.
Gilbert, S. F. 1991. Developmental Biology. MA: Sinauer Associates Inc., 3rd edition.
Mange, D., and Tomassini, M., eds. 1998. Bio-inspired
Computing Machines: Towards Novel Computational
Architectures. Lausanne, Switzerland: Presses Polytechniques et Universitaires Romandes.
Mange, D.; Sanchez, E.; Stauffer, A.; Tempesti, G.; Marchal, P.; and Piguet, C. 1998. Embryonics: A new
methodology for designing field-programmable gate
arrays with self-repair and self-replicating properties.
IEEE Transactions on VLSI Systems 6(3):387–399.
May, M. 2000. Mother nature’s menders. the battle against aging. Scientific American. Special Issue
”The Quest to Beat Aging”.
Negrini, R.; Sami, M. G.; and Stefanelli, R. 1989. Fault
Tolerance Through Reconfiguration in VLSI and WSI
Arrays. Cambridge, MA: MIT Press.
Ortega, C., and Tyrrell, A. 1999. Reliability analysis in self-repairing embryonic systems. In Proc. 1st
NASA/DoD Workshop on Evolvable Hardware, 120–
128.
Prodan, L.; Tempesti, G.; Mange, D.; and Stauffer, A.
2000. Biology meets electronics: The path to a bioinpired FPGA. In Proc. 3rd International Conference on Evolvable Systems: From Biology to Hardware, 187–196.
Prodan, L.; Tempesti, G.; Mange, D.; and Stauffer, A.
2001. Embryonics: Artificial cells driven by artificial
DNA. In Proc. 4th International Conference on Evolvable Systems: From Biology to Hardware, 100–111.
Shamblott, M., et al. 1998. Derivation of pluripotent
stem cells from cultured human primordial germ cells.
Proc. Natl. Acad. Sci. USA 95:13726–13731.
Shibayama, A.; Igura, H.; Mizuno, M.; and Yamashina,
M. 1997. An autonomous reconfigurable cell array for
fault-tolerant LSIs. In Proc. 44th IEEE International
Solid-State Circuits Conference, 230–231.
Sipper, M.; Sanchez, E.; Mange, D.; Tomassini, M.;
Perez-Uribe, A.; and Stauffer, A. 1997. A phylogenetic, ontogenetic, and epigenetic view of bio-inspired
hardware systems. IEEE Transactions on Evolutionary Computation 1(1):83–97.
Sipper, M.; Tempesti, G.; Mange, D.; and Sanchez, E.
1998. Von Neumann’s legacy: Special issue on self-
in Artificial Life VIII, Standish, Abbass, Bedau (eds)(MIT Press) 2002. pp 101–105
replication. Artificial Life 4(3).
Sipper, M. 1998. Fifty years of research on selfreplication: an overview. Artificial Life 4(3):237–257.
Tempesti, G.; Mange, D.; and Stauffer, A. 1998. Selfreplicating and self-repairing multicellular automata.
Artificial Life 4(3):259–282.
Tempesti, G. 1998. A Self-Repairing Multiplexer-Based
FPGA Inspired by Biological Processes. Ph.D. Dissertation, EPFL, Lausanne. No. 1827.
Thomson, J., et al. 1998. Embryonic stem cell lines
derived from human blastocysts. Science 282:1145–
1147.
Vaughan, C. 2000. Stem cells’ “guardians” found to
control cell specialization. Stanford Report.
Wolpert, L. 1991. The Triumph of the Embryo. New
York: Oxford University Press.
5