1 Embryomorphic Embryomorphic Engineering: Engineering: From biological development to From biological development to self self - - organized organized computational computational architectures architectures René Doursat http://www.iscpif.fr/~doursat
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Self-organized systems that look like they were designed
1. What are Complex Systems?•
Decentralization•
Emergence•
Self-organization
2. Architects Overtaken by their Architecture Designed systems that became suddenly complex
4. Embryomorphic EngineeringFrom biological cells to robots and networks
5. The New Challenge of "Meta-Design"
Or how to organize spontaneity
but were not
5
Complex systems can be found everywhere around us
1. What are Complex Systems?1. What are Complex Systems?
Internet& Web
= host/page
insectcolonies
= ant
patternformation
= matter
biologicaldevelopment
= cell
socialnetworks= person
the brain& cognition
= neuron
Physical, biological, technological, social complex systems
a) decentralization: the system is made of myriads of "simple" agents (local information, local rules, local interactions)
b) emergence: function is a bottom-up collective effect of the agents (asynchrony, balance, combinatorial creativity)
c) self-organization: the system operates and changes on its own (autonomy, robustness, adaptation)
6
Mammal fur, seashells, and insect wings(Scott Camazine, http://www.scottcamazine.com)
NetLogo Fur simulation
Ex: Pattern formation – Animal colorsanimal patterns caused by pigment cells that try to copy their nearest neighbors but differentiate from farther cells
1. What are Complex Systems?1. What are Complex Systems?
Ex: Swarm intelligence – Insect coloniestrails form by ants that follow and reinforce each other’s pheromone path
Harvester ants (Deborah Gordon, Stanford University)
clusters and cliques of people who aggregate in geographical or social spacecellular automata model
8
1. What are Complex Systems?1. What are Complex Systems?All kinds of agents: molecules, cells, animals, humans & technology
the brain organisms ant trails
termitemounds
animalflocks
cities,populations
social networksmarkets,economy
Internet,Web
physicalpatterns
living cell
biologicalpatterns cells
animals
humans& tech
molecules
9
a) Decentralization: the system is made of myriads of "simple" agentslocal information (no group-level knowledge): each agent carries a piece of the global system’s statelocal rules (no group-level goals): each agent follows an individual agendalocal interactions (no group-level scope): each agent communicates with "neighboring" agents, possibly via long-range links
b) Emergence: function is a bottom-up collective effect of the agentsasynchronous dependencies: agents "threaded" in parallel modify each other’s actions (possibly via cues they leave in the environment)balance: creation by +feedback (imitation), control by –feedback (inhibition)combinatorial creativity: the system exhibits new (surprising) properties that the agents do not have; different properties can emerge fromthe same agents
1. What are Complex Systems?1. What are Complex Systems?3 main differences with traditional architecting
10
c) Self-organization: the system operates and changes on its ownautonomy: there is no external map, grand architect, or explicit leaderrobustness: proper function is maintained despite (some) damageadaptation: the system dynamically and "optimally" varies with a changing environment; agents modify themselves to create a new class of functional collective behaviors → learning and/or evolution
1. What are Complex Systems?1. What are Complex Systems?3 main differences with traditional architecting
•
decentralized, emergent, self-organized processes are the rule in nature and large-scale human superstructures
•
however, they are counterintuitive to our human mind, which prefers central-causal, predictable, planned/rigid systems
•
... and yet again, autonomy, robustness, adaptation are highly desirable properties! How can we have it both ways, i.e. "care and let go"?
11
Paris Ile-de-France
LyonRhône-Alpes
National
4th French Complex SystemsSummer School, 2010
1. What are Complex Systems?1. What are Complex Systems?
12
A vast archipelago of precursor and neighboring disciplines
dynamics: behavior and activity of a system over time multitude, statistics: large-scale
properties of systems
adaptation: change in typical functional regime of a system
complexity: measuring the length to describe, time to build, or resources to run, a system
dynamics: behavior and activity of a system over time
Self-organized systems that look like they were designed
1. What are Complex Systems?•
Decentralization•
Emergence•
Self-organization
2. Architects Overtaken by their Architecture Designed systems that became suddenly complex
4. Embryomorphic EngineeringFrom biological cells to robots and networks
5. The New Challenge of "Meta-Design"
Or how to organize spontaneity
but were not
Complex systems seem so different from architected systems, and yet...
14
cities,populations
Internet,Web social networksmarkets,
economy
companies,institutions
addressbooks
houses,buildings
computers,routers
2. Architects Overtaken by their Architecture2. Architects Overtaken by their ArchitectureAt large scales, human superstructures are "natural" CS
... arising from a multitude of traditionally designed artifactshouses, buildings
address books
companies, institutions
computers, routers
large-scaleemergence
small to mid- scale artifacts
by their unplanned, spontaneous emergence and adaptivity...
geography: cities, populationspeople: social networks
wealth: markets, economytechnology: Internet, Web
15
2. Architects Overtaken by their Architecture2. Architects Overtaken by their ArchitectureAt mid-scales, human artifacts are classically architected
a goal-oriented, top-down process toward one solution behaving in a limited # of ways
specification & design: hierarchical view of the entire system, exact placement of eltstesting & validation: controllability, reliability, predictability, optimality A
rchi
Mat
e EA
exa
mpl
e
the (very) "complicated" systems of classical engineering and social centralization
electronics, machinery, aviation, civil construction, etc.spectators, orchestras, administrations, military (reacting to external cues/leader/plan)
not "complex" systems:little/no decentralization, little/no emergence, little/no self-organization
New inflation: artifacts/orgs made of a huge number of parts
Syst
ems
engi
neer
ing
Wik
imed
ia C
omm
ons
16
number of transistors/year
in hardware, software,agents, objects, services
number of O/S lines of code/year
networks...
number of network hosts/year
Burst to large scale: de facto complexification of ICT systemsineluctable breakup into, and proliferation of, modules/components
2. Architects Overtaken by their Architecture2. Architects Overtaken by their Architecture
→
trying to keep the lid on complexity won’t work in these systems:cannot place every part anymorecannot foresee every event anymorecannot control every process anymore ... but do we still want to?
17
Large-scale: de facto complexification of organizations, via techno-social networks
ubiquitous ICT capabilities connect people and infrastructure inunprecedented waysgiving rise to complex techno-social "ecosystems" composed of a multitude of human users and computing devices
2. Architects Overtaken by their Architecture2. Architects Overtaken by their Architecture
→ in this context, impossible to assign every single participant a predetermined role
healthcare energy & environmenteducation defense & securitybusiness finance
from a centralized oligarchy of providers ofdata, knowledge, management, information, energy
to a dense heterarchy of proactive participants:patients, students, employees, users, consumers, etc.
explosion in size and complexity in all domains of society:
18
a) Overtakenhow things turned around from top-down "architecting as usual" (at mid scales) and went bottom-up (at large-scales)⎯hopefully not yet belly-uplarge-scale techno-social systems exhibit spontaneous collective behavior that we don’t quite understand or control yet
b) Embracethey also open the door to entirely new forms of enterprise characterized by increasing decentralization, emergence, and dynamic adaptation
c) Take overthus it is time to design new collaborative technologies to harness and guide this natural (and unavoidable) force of self-organizationtry to focus on the agents’ potential for self-assembly, not the system
→ 4. Embryomorphic Engineering → 5. "Meta-Design"
2. Architects Overtaken by their Architecture2. Architects Overtaken by their ArchitectureThe "New Deal" of the ICT age
Ex: Morphogenesis – Biological developmentcells build sophisticated organisms by division, genetic differentiation and biomechanical self-assemblyNadine Peyriéras, Paul Bourgine et al.
(Embryomics & BioEmergences)www.infovisual.info
architecture
23
3. Architecture Without Architects3. Architecture Without ArchitectsFrom “statistical” to “morphological” CS
in inert matter / insect constructions / multicellular organisms
ant trail
network of ant trails
ant nesttermite mound
more intrinsic, sophisticated architecturephysicalpattern formation
biologicalmorphogenesis
social insectconstructions
grains of sand + air insectscells
new inspiration
24
Complex systems can possess a strong architecture, too
self-reconfiguring manufacturing plantself-stabilizing energy gridself-deploying emergency taskforce
. . . self-architecting enterprise?
27
3. Architecture Without Architects: ICT-like CSSome natural complex systems strikingly demonstrate the possibility of combining pure self-organization and elaborate architectures
→ how can we extract and transfer their principles to human artifacts⎯ such as EA?
2. Architects Overtaken by their Architecture: CS-like ICTConversely, mid- to large-scale techno-social systems already exhibit complex systems effects⎯albeit still uncontrolled and, for most of them, unwanted at this point
→ how can we regain (relative) control over these "golems"?
RECAP Toward a reconciliation of complex systems and ICT
3. Architecture Without Architects3. Architecture Without Architects
change of signals,chemical messengersdiffusion gradients
("morphogens")
A closer look at morphogenesis: ⇔ it couples mechanics and genetics4. 4. EmbryomorphicEmbryomorphic EngineeringEngineering
33
grad1
div1
patt1
div2
grad2
patt2div3
grad3
patt3...
Alternation of self-positioning (div)and self-identifying(grad/patt)
genotype
4. 4. EmbryomorphicEmbryomorphic EngineeringEngineeringCapturing the essence of morphogenesis in an Artificial Life agent model
each agentfollows the same setof self-architecting rules (the "genotype")but reacts differently depending on its neighbors Doursat (2009)
18th GECCO
34r
pA
BV
rr0rerc
div
GSA: rc < re = 1 << r0p = 0.05
35
grad
EW
S
N
EW
WE WENS
36
I4 I6
B4
B3
patt
X Y
. . . I3 I4 I5 . . .
B1 B2 B4B3
wix,iy
GPF : {w }
wki
WE NS
37
I9
I1
(a) (b)
(c)
. . . . . .
WE = X NS = Y
B1 B2 B3 B4
I3 I4 I5
X Y
. . . I3 I4 I5 . . .
B1 B2 B4B3
wiX,YGPF
wki
Programmed patterning (patt): the hidden embryo atlasa) same swarm in different colormaps to visualize the agents’ internal
patterning variables X, Y, Bi and Ik (virtual in situ hybridization)b) consolidated view of all identity regions Ik for k = 1...9c) gene regulatory network used by each agent to calculate its expression
a) Giving agents self-identifying and self-positioning abilitiesagents possess the same set of rules but execute different subsets depending on their position = "differentiation" in cells, "stigmergy" in insects
b) ME brings a new focus on "complex systems engineering"exploring the artificial design and implementation of autonomous systems capable of developing sophisticated, heterogeneous morphologies or architectures without central planning or external lead
ME is about programming the agents of emergence
swarm robotics,modular/reconfigurable roboticsmobile ad hoc networks,sensor-actuator networkssynthetic biology, etc.
c) Related emerging ICT disciplines and application domainsamorphous/spatial computing (MIT)organic computing (DFG, Germany)pervasive adaptation (FET, EU)ubiquitous computing (PARC)programmable matter (CMU)
Self-organized systems that look like they were designed
1. What are Complex Systems?•
Decentralization•
Emergence•
Self-organization
2. Architects Overtaken by their Architecture Designed systems that became suddenly complex
4. Embryomorphic EngineeringFrom biological cells to robots and networks
5. The New Challenge of "Meta-Design"
Or how to organize spontaneity
but were not
52
ME and other emerging ICT fields are all proponents of the shift from design to "meta-design"
www.infovisual.info
fact: organisms endogenously grow but artificial systems are builtexogenously
challenge: can architects "step back" from their creation and only set the generic conditions for systems to self-assemble?
instead of building the system from the top ("phenotype"),program the components from the bottom ("genotype")
systems designsystems"meta-design"
genetic engineering
direct (explicit)
indirect (implicit)
5. The New Challenge of 5. The New Challenge of "Meta"Meta--Design"Design"
53
Between natural and engineered emergence
CS engineering: creating and programminga new "artificial" emergence
→ Multi-Agent Systems (MAS)
CS science: observing and understanding "natural", spontaneous emergence (including human-caused)
→ Agent-Based Modeling (ABM)
CS meta-design: fostering and guiding complex systems (e.g. techno-social)
5. The New Challenge of 5. The New Challenge of "Meta"Meta--Design"Design"
But CS meta-design is notwithout its paradoxes...• Can we plan their
autonomy? • Can we control their
decentralization?• Can we program their
adaptation?
54
People: the ABM modeling perspective of the social sciencesagent- (or individual-) based modeling (ABM) arose from the need to model systems that were too complex for analytical descriptionsmain origin: cellular automata (CA)
von Neumann self-replicating machines → Ulam’s "paper" abstraction into CAs → Conway’s Game of Lifebased on grid topology
other origins rooted in economics and social sciencesrelated to "methodological individualism"mostly based on grid and network topologies
Macal & NorthArgonne National Laboratory
later: extended to ecology, biology and physicsbased on grid, network and 2D/3D Euclidean topologies
→
the rise of fast computing made ABM a practical tool
5. The New Challenge of 5. The New Challenge of "Meta"Meta--Design"Design"
55
ICT: the MAS multi-agent perspective of computer scienceemphasis on software agent as a proxy representing human users and their interests; users state their prefs, agents try to satisfy them
ex: internet agents searching informationex: electronic broker agents competing / cooperating to reach an agreementex: automation agents controlling and monitoring devices
main tasks of MAS programming: agent design and society designan agent can be ± reactive, proactive, deliberative, socialan agent is caught between (a) its own (sophisticated) goals and (b) the constraints from the environment and exchanges with the other agents
→
meta-design should blend both MAS and ABM philosophiesMAS: a few "heavy-weight" (big program), "selfish", intelligent agentsABM: many "light-weight" (few rules), highly "social", "simple" agentsMAS: focus on game theoretic gainsABM: focus on collective emergent behavior
5. The New Challenge of 5. The New Challenge of "Meta"Meta--Design"Design"
56
a) Construe systems as self-organizing building-block gamesInstead of assembling a construction yourself, shape its building blocks in a way that they self-assemble for you—and come up with new solutions
TAKEAWAY Getting ready to organize spontaneity
5. The New Challenge of 5. The New Challenge of "Meta"Meta--Design"Design"
b) Design and program the piecestheir potential to search, connect to, interact with each other, and react to their environment
c) Add evolutionby variation (mutation) of the pieces’ program and selection of the emerging architecture
mut
atio
n
mut
atio
n
mut
atio
n
51208
10392
117614
47213
30 434 422316
2210210
differentiation
• piece = "genotype"• architecture = "phenotype"
57
4. Embryomorphic EngineeringFrom biological cells to robots and networks