I N T E L L I G E N C E R

Features Editor: Crystal R. Chweh, [email protected]

• Alife attempts to simulate evolution

• Data mining e-business

• NASA robotics could help patients walk

Artificial intelligence experiments that the scientific method says

we must do—but cannot with the time

Life as it could be: and/or spatial scales of material structures

like cells themselves.”
Alife purports to shed light on our

Alife attempts to simulate understanding of “life as it could be” by

synthesizing it in artificial media, in the

evolution same way that chemical synthesis aided

chemists in developing a fundamental set
of laws in that field. A century ago, chem-
By Nancy Forbes
ists’ attempts to understand how matter
behaved were limited to what they could
glean from the analysis of chemical reac-
Artificial life explores natural life by attempting to create tions they performed on readily available
substances. “To have a theory of the ac-
biological phenomena in computers and other nonliving tual, it is necessary to understand the pos-
sible,” remarked Chris Langton, one of the
media. Alife also strives to introduce a broader, more universal founders of Alife, in the introduction to his
understanding of life—accessible to new kinds of manipula- first book on the subject.1 However, by
developing the capability to synthesize
tion and testing—than that defined by traditional biology. new chemicals not found in nature, scien-
tists have not only greatly extended their
Looked at in this way, Alife represents would inevitably have to develop this way. empirical basis for study and their under-
the flip side of the empirical–analytic ap- Couldn’t silicon-based life or germanium- standing of underlying theory, but have
proach of most laboratory biologists, who based life exist somewhere? While provid- also produced substances such as plastic,
start with a whole organism, deconstruct it ing considerable fodder for science fiction rubber, and certain pharmaceuticals that
into its component parts (for example or- writers, this question has arisen for tradi- have proved extremely useful to society.
gans, tissues, cells, genes, molecules, and tional evolutionary biologists and Alife Alife proponents generally consider the
so on), and then try to derive its fundamen- researchers alike. According to the latter, field to have arisen from the initial efforts of
tal principles. Most biologists would prefer the study of carbon-based life here on earth Langton, an autodidact who was preoccu-
to be able to simply reverse engineer life constitutes a special case and gives neces- pied with these questions throughout the
and synthesize real organisms, cookbook sary but not sufficient information for for-
style, from their basic ingredients, but so mulating the basic principles of life as
far, this feat has eluded them. shared by all living systems on all planets.
To uncover these principles, they feel, it
Non-carbon-based life? would be necessary to explore the space of
Biology is the exploration of life on all possible biologies, by studying many
earth. Despite its vast complexity and different life forms. But until aliens show
diversity this remarkable phenomenon up here on earth for this purpose, most Alife
given to us by nature nonetheless results scientists feel that the artificial simulation
from a series of historical accidents that of life can bring us closer to that goal.
occurred through a causal chain of events. Says the Santa Fe Institute’s John Casti:
Although the production of life on earth “I think Alife will ultimately enable us to Figure 1: A picture of a Golem project’s computer-
followed a path based on carbon-chain properly understand evolution and the work- designed evolvable robot, which pushes itself along
chemistry, some scientists legitimately ings of cellular machinery, mostly because it the carpet using the piston at the center. (Source: The
question whether, on any other planet, life will offer us the chance to do the kinds of Golem Project at Brandeis University).

2 IEEE INTELLIGENT SYSTEMS

 Synapse
Neuron
Linear actuator

seventies. The first workshop on Alife (the

Interdisciplinary Workshop on the Synthe-
sis and Simulation of Artificial Life), spon- Bar
Control (brain)
sored by Los Alamos National Laboratory,
the Santa Fe Institute, and Apple Computer, Morphology
officially gave “birth” to the field in Sep- Ball joint (body)
tember 1987.2 You can find, however, pre-
cursors for this field in both Turing’s and Infinite plane
von Neumann’s work on automata.

Alife for individual entities

Alife encompasses many disciplines, such
as engineering, computer science, biology,
physics, chemistry, sociology, and even eco-
nomics. In his textbook on subject, Caltech’s
Chris Adami defines Alife research as a con- Figure 2. A schematic illustration of The Golem project’s evolvable robot. Bars and actuators were used as building
tinuum with the simulation–emulation of blocks of structure, and artificial neurons as building blocks of control. Bars connected with free joints can potentially
individual entities at one end. As an example, form trusses that represent arbitrary rigid, flexible, and articulated structures and emulate revolute, linear, and planar
he gives Karl Sims’s simulation of the evolu- joints at various levels of hierarchy. Similarly, sigmoidal neurons can connect to create arbitrary control architectures
tion of the form and movements of swim- such as feed-forward and recurrent nets, state machines, and multiple independent controllers. Neurons were allowed
ming behavior in virtual animals made from to connect to bars, similar to real neurons that govern the contraction of muscle tissue.
blocks, and how competitive behavior devel-
ops in these creatures.3
An analogous example in engineering Polluck, the parent bots consist of a com- each wasp really “knowing” what it’s build-
would be the construction of adaptive, puter running an evolutionary algorithm ing.7 Craig Reynolds’ work on flocking birds,
autonomous robots, which can interact that produces a design based on trial and offers another example. Reynolds created a
with their environment, evolve, and learn error and a 3D printer that makes small virtual flock of birds, called boids, which
from it—as exemplified by the robotic plastic shapes. The offspring are small flew according to three rules:
crickets developed by Barbara Web’s group plastic trusses with actuators, propelled by • Always avoid collisions with your neighbors.
at the University of Edinburgh.4 This motors and controlled by artificial neural • Always try to fly at the same speed as your
method takes a bottom-up approach, in nets (see Figures 1 and 2). Humans inter- neighbors.
which developers construct a robotic sys- vene only to attach the motors and connect • Always try to stay close to your neighbors.
tem from simple elemental units that— the wires—the robots do all the rest,
through evolution, emergence, and adapta- including telling the humans what to do. These three rules sufficed to create the
tion to their environments—expand into emergence of flocking behavior. The boids
more complex systems. Emergent behavior flew as a coherent group and automatically
Robots constructed with artificial intelli- At the other end of the Alife spectrum, split into two groups when encountering an
gence, on the other hand, employ a top- Adami places the study of the emergent prop- obstacle, reuniting after it was passed
down approach, where developers start by erties of living populations, which display through. This system demonstrates how a
targeting a complex behavior (such as properties that can’t be seen in the individual system of fairly simple elements—interact-
walking up steps) and build the system units’ behavior. (For example, in physics, ing with their nearest neighbors, with no
with all the elements it needs to achieve temperature and pressure exemplify emer- central direction—can create cohesive, intel-
this behavior. What’s more, AI has tradi- gent behaviors that occur in large systems of ligent group behavior. In fact, the behavior
tionally focused on machines achieving interacting molecules. An individual mole- of Reynold’s boids precisely emulates what
complex, multifaceted human functions, cule has neither temperature nor pressure by we observe in nature.
such as chess playing, voice comprehen- itself.) According to Adami, these living sys-
sion, or medical diagnosis. Conversely, tems aren’t amenable to a statistical descrip- Life on Tierra
Alife looks exclusively at natural behav- tion in terms of macroscopic variables, so he Somewhere in between these two extremes
iors, emphasizing survivability, evolution, substitutes describing them in parallel to lies Tom Ray’s Tierra project, which seeks to
and reproduction of the “creature” in com- show emergent behavior. However, even the explore open-ended evolution in a virtual
plex, dynamic environments. latter approach falls short because parallelism world, unfolding without any a priori human
The Golem (genetically organized life- can’t capture the self-organization seen in instructions. Ray modeled Tierra on the per-
like electro-mechanics) project at Brandeis many living systems and because the result- iod in Earth’s evolution known as the Cam-
University presents an even more fascinat- ing emergent behavior can often affect its brian Era, about 600 million years ago. The
ing example of engineered virtual life, members in nonlinear ways.6 period began with the existence of simple,
involving robots that can actually design Examples of this include swarm intelli- self-replicating organisms, which underwent
and build other robots.5 Devised by com- gence that emerges, for example, when wasps explosive growth over time to result in the
puter scientists Hod Lipson and Jordan build large structures, such as nests, without great species diversity known today.

NOVEMBER/DECEMBER 2000 3
 IEEE

& their applications

IEEE Computer Society

Publications Office
10662 Los Vaqueros Circle, PO Box 3014
Los Alamitos, CA 90720-1314

STAFF
Group Managing Editor
Dick Price
[email protected]
Associate Editor
Dennis Taylor
[email protected]
Managing Editor
Dale Strok
Features Editor
Crystal Chweh
Staff Editors
Cheryl Baltes, Jenny Ferrero, and Shani Murray
Magazine Assistants
80 45 81 17 140 78 82 125 28 others
0.0500 0.0472 0.0181 0.0167 0.0139 0.0043 0.0049 0.0047 0.0049 0.0250
Dawn Craig and Angela Williams
Contributing Editors Figure. 3: An evolutionary race between hosts and parasites in a primordial soup of the Tierra program. Each image
Nancy Forbes, Marti Hearst, represents a soup of 60,000 bytes, divided into 60 segments of 1,000 bytes each. A colored bar represents each indi-
Paula Powers, Keri Schreiner, and Gil Shif vidual creature. Colors correspond to genome size (for example, red = 80, yellow = 45, blue = 79). In this image,
Design Director
hosts (red) are very common. Parasites (yellow) have also appeared but are still rare. (Photo: Marc Cygnus)
Toni Van Buskirk
Layout/Technical Illustrations Ray wanted to investigate how self-repli- tion, and so on—killing organisms when
Alex Torres and Carmen Flores-Garvey they reach the top of the queue; and
cation eventually produced such complex,
varied life forms. He started with a single • doling out random mutations in the binary
Interim Executive Director
Anne Marie Kelly string of each organism’s program, thus
organism called the Ancestor, the only engi-
causing some organisms to self-replicate
neered life form in Tierra. He then let the
Publisher imperfectly.8
Angela Burgess creature loose and watched to see what hap-
pened. After only one night, his virtual Ray, currently a zoology and computer
Membership/Circulation Marketing Manager
Georgann Carter
world was teeming with myriad creatures, science professor at the University of Okla-
displaying an amazing variety of form and homa, first attempted Tierra in January
Advertising Staff
Sandy Aijala, Marian Anderson, and Debbie Sims behavior. These organisms and their prog- 1990. After over 500 million instructions,
eny (the “organisms” are actually self-repro- Tierra had created over 350 different sizes
Submissions: Submit two copies of all articles and track proposals
to Intelligent Systems, Magazine Assistant, 10662 Los Vaqueros
ducing programs written in assembler lan- of life forms, 93 of which had survived to
Circle, Los Alamitos, CA 90720-1314, phone +1 714 821 8380, guage) competed for the natural resources of achieve subpopulations of five or more
[email protected]. Manuscripts should be approximately
5,000 words long, preferably not exceeding 10 references. Visit
their world—that is, CPU time and memory. individuals. Tierra also generated hosts and
computer.org/intelligent for editorial guidelines. This provided the basis for natural selection parasites, and eventually a type of social
Editorial: Unless otherwise stated, bylined articles as well as to operate—some organisms die off and the organization with communities of geneti-
products and services reflect the author’s or firm’s opinion; inclusion fitter ones survive, adapt, and become more cally uniform organisms. Nearly every
does not necessarily constitute endorsement by the IEEE Computer
Society or the IEEE. competitive. facet of natural evolution and known life-
Circulation: IEEE Intelligent Systems (ISSN 1094-7167) is To prevent these digital beings from form behavior showed its face in Tierra,
published bimonthly by the IEEE Computer Society. IEEE gaining access to the actual hardware of including competitive, exploitative, and
Headquarters, Three Park Ave., 17th Floor, New York, NY 10016-
5997; IEEE Computer Society Publications Office, 10662 Los the machine they lived in like computer protective behaviors.
Vaqueros Circle, PO Box 3014, Los Alamitos, CA 90720-1314, viruses, Ray made the entire Tierra pro-
phone +1 714 821 8380; IEEE Computer Society Headquarters, 1730
Massachusetts Ave. NW, Washington, DC 20036-1903. Subscription gram run on a virtual computer created in Life among the Avidians
rates: IEEE Computer Society members get the lowest rates and the software. Tierra’s operating system Inspired by Ray’s work, Adami and
choice of media option – $32/26/42 US print/electronic/combination.
Go to http://computer.org/subscribe to order and for more information basically performed four functions: Richard Lenski, a microbiologist at Michi-
on other subscription prices. Nonmember rate: available on request.
gan State University, began their own Alife
Back issues: $10 for members and $20 for nonmembers. This • allocation of memory to each organism, experiments a few years later. Lenski had
magazine is available on microfiche.
letting the organism have the exclusive privi-
Copyright and reprint permissions: Abstracting is permitted been conducting (and continues to conduct)
with credit to the source. Libraries are permitted to photocopy
lege of modifying its own structure, to pre-
serve its unique identity;
wet lab experiments on evolution in his lab
beyond the limits of US copyright law for the private use of patrons
1) those post-1977 articles that carry a code at the bottom of the • allocation of CPU time to each organism so with E. coli bacteria, where a single experi-
first page, provided the per-copy fee indicated in the code is paid ment can span up to 24,000 generations—
through the Copyright Clearance Center, 222 Rosewood Drive, it could act;
Danvers, MA 01923; and 2) pre-1978 articles without fee. For other • placement of each organism in a queue usually, a new one about every 3.5 hours.9
copying, reprint, or republication permissions, write to the Copyright
and Permissions Department, IEEE Service Center, 445 Hoes Lane,
and—depending on life cycle, natural selec- After having read Adami’s book, Lenski
Piscataway, NJ 08855-1331. Copyright © 2000 The Institute of Electrical
and Electronics Engineers, Inc. All rights reserved.
IEEE INTELLIGENT SYSTEMS
 How to Reach Us
Writers
For detailed information on submitting articles,
write for our Editorial Guidelines (isystems@
computer.org), or access computer.org/
intelligent/edguide.htm.

Letters to the Editor

Send letters to
Dennis Taylor
Associate Editor
IEEE Intelligent Systems
10662 Los Vaqueros Circle
Los Alamitos, CA 90720
[email protected]
Please provide an e-mail address or
daytime phone number with your letter.
Figure 4. Example of an Avida lattice during a computer run at a reasonably low mutation rate. A different color is On the Web
assigned each genotype that has more than three organisms of that type. Genotypes that don’t reach this threshold
are colored dark blue. The Avida world shown is toroidal, so each edge wrap is connected to the opposite edge of the Access computer.org/intelligent for information
figure. This picture was taken at update 3,100 of an 100 × 100 grid. about IEEE Intelligent Systems.

Subscription Change of Address

soon contacted him to learn more about virtual results and actual ones on the same Send change-of-address requests for magazine
Alife, and Avida—a new software program footing, to test how accurately Alife can subscriptions to [email protected]. Be
developed by Titus Brown and Charles replicate the workings of natural evolution. sure to specify Intelligent Systems.
Ofria, Adami’s students—emerged. However, Princeton evolutionary biolo-
Avida would not only enable digital life gist Rob Knight feels this comparison Membership Change of Address
forms to evolve, but would subject them to requires more subtle treatment: “A major Send change-of-address requests for the mem-
experiments in evolution, such as inserting distinction in the Alife community is be-
bership directory to directory.updates@
a particular type of mutation and seeing tween weak and strong Alife. Weak Alife,
what resulted in the ensuing generations. which claims that simulations of evolving
computer.org.
The program typically ran for about 5,000 systems may help us understand biological
Missing or Damaged Copies
to 20,000 generations, taking only two to life, is relatively uncontroversial, especially
eight hours to complete. By comparison, when the simulations relate closely to nat-
If you are missing an issue or you received
similar experiments performed in vivo or ural systems, such as Nilsson and Pelger’s a damaged copy, contact membership@
in vitro in a lab could take days. study of the evolution of eye morphology.10 computer.org.
Together, Adami and Lenski created a These [simulations] clearly provide valuable
virtual environment where one initial bug insight. However, experimental biologists Reprints of Articles
was programmed to reproduce itself and tend to distrust computer experiments, per- For price information or to order reprints, send e-
mutate every 1,000 or so “births.” The bug haps unfairly, on the grounds that models mail to [email protected] or fax +1 714
could also perform simple mathematical often merely reflect biases programmed in at 821 4010.
functions that the environment rewarded, by the start. Strong Alife, which claims that
allowing those bugs to replicate at a faster replicating programs inside computers really Reprint Permission
rate. This helped push out the less compe- are alive, is far more controversial. This is To obtain permission to reprint an article, contact
tent bugs, forcing them to eventually die partly because the examples of biological William Hagen, IEEE Copyrights and Trademarks
off—emulating natural selection. life we are all familiar with are orders of Manager, at [email protected].
“Today,” says Adami, “we try to do exper- magnitude more complex and partly because
iments with both our virtual Avidians and E. the claimed similarities with biological evo-
coli in vitro to be able to compare the re- lution tend to be rather abstract.”
sults.” The researchers have found that the IEEE
evolutionary principles seen in the comput- Alife vs. life itself
erized environment accurately mirror those The question of the nature and origin of
found in nature, and that the “survivors” in life has preoccupied human thought since
their system appeared stronger and less af- the dawn of time. It has been answered in
fected by random mutations than less fit in- turn by priests, philosophers, shamans,
dividuals. The experiment purports to put seers, mystics, scientists, and charlatans.
& their applications
NOVEMBER/DECEMBER 2000
 Data mining

Data mining ment infrastructure of a fictional e-commerce company, includ-

ing a Web server, an industry-strength database system, Web-

e-business
database connectivity, and applications such as data mining.”
Students will use the lab to construct and use large databases
as well as to design and conduct Web-based surveys and primary
Cornell University’s Johnson Graduate School of Manage- market research for business clients. They will also develop mod-
ment (www.johnson.cornell.edu) has received a $200,000 els to attract more visitors to Web sites, keep them there longer,
grant from Intel to create an e-business database management improve look-to-buy ratios, retain profitable customers, learn
laboratory. As part of a new program in e-business studies, about customers’ preferences, and personalize marketing offers.
students will use the lab to do their own data mining and inter- “Cornell will be one of the first schools to have a laboratory
active analysis of customers’ behavior generated by Web- specifically designed to introduce students to this technology
based businesses. When the Johnson School’s immersion and to allow faculty to conduct research with contemporary
course on e-commerce debuts in January 2001, the lab will be resources,” says Richard Conway, leader of the e-business
fully equipped with workstations, servers, networking equip- immersion faculty team and the Emerson Electric Professor
ment, and software. emeritus at the Johnson School.
“The lab will allow us to offer a hands-on approach to learn- Cornell is one of five universities worldwide selected to receive
ing about e-business,” says Johannes Gehrke, assistant profes- Intel grants supporting e-business studies this year. The others are
sor of computer science in Cornell’s Faculty of Computing and Carnegie Mellon, Harvard, and the University of Michigan in the
Information Sciences group. “It will support a data manage- United States and Tsinghua University in China.

All the religions and all the mythologies of looked at as a simulacrum for real biology, life is somewhat specious. It is frequently
the world have given it a central place in many mainstream biologists feel it does not difficult to draw meaningful scientific dis-
their belief systems, and only lately in the offer a legitimate vehicle for the study of tinctions between organisms, viruses, and
history of mankind has it become the pur- living organisms or how they evolved. growing crystals. Thus, I have no problem
view of science—creationism notwith- When asked for his views on Alife, chemist with those who say that life inside a com-
standing. Given this import, it’s not surpris- Stanley Miller, who in 1952 conducted the puter is ‘real’ life; the word is as ambiguous
ing that the idea of computers simulating first experiment that simulated the primor- inside a computer as outside. From this
life should elicit debate, controversy, and dial soup of the primitive earth and pro- standpoint, while creatures (or ‘replicators’)
even incredulity. duced amino acids, retorted, “Running spawned by Alife couldn’t adapt to a biologi-
On one hand, Alife represents a dynamic, equations through a computer does not con- cal environment (imagine cellular automata
new multidisciplinary field that continues to stitute an experiment!”11 However, judging in a rain forest), they can compete and adapt
thrive and attract new followers. It has pro- the merits of Alife solely on the basis of in a virtual one. So artificial life shouldn’t be
duced some remarkable research, and its how well its answers stack up against those judged on the basis of whether or not it’s as
scientists are making inroads in a plethora derived from traditional biology could dis- valid as biological life, but should be re-
of disciplines, such as algorithm develop- tort and even diminish its accomplishments. garded as a completely separate entity, which
ment, software, hardware, robotics, popula- Andy Ellington, a biochemist and engi- a priori doesn’t need to have the same under-
tion biology, economics, and complex sys- neer who studies evolution at the University lying rules as biological replicators. What we
tems analysis. of Texas at Austin believes the problem need is a paradigm shift in the way we think
However, some scientists view Alife with might lie with confusion over how we define about the concept of life.”
wariness, skepticism, and dismissal. When life. Says Ellington, “The whole notion of Rob Knight agrees: “To be convincing,
future Alife research needs to clearly define
life and its characteristics, and show that
Alife experiments can both recapture and
For further reading extend the results of more traditional meth-
ods of inference about evolution.”
Alife Bibliography: www.cogs.susx.ac.uk/users/ezequiel/alife-page/alife.html However it’s viewed, Alife has not only
The Alife Database: http://arieldolan.com/aldb/default.asp spawned some fascinating varieties of digi-
Alife VII Conference: http://alife7.alife.org
tal life, but also an exciting new field of
Artificial Life (MIT Press): http://mitpress.mit.edu/journal-home.tcl?issn=10645462
research. Moreover, Alife has engendered a
Artificial Life Online 2.0: http://alife.org
Boids Page: www.red3d.com/cwr/boids rich and provocative body of thought—
Digital Life Laboratory: http://dllab.caltech.edu whether you’re for or against it—with the
The Ellington Lab at the University of Texas at Austin: http://ellingtonlab.org potential for generating an endless number
The Golem Project: http://golem03.cs-i.brandeis.edu/index.html of new ideas.
Technosphere: www.technosphere.org.uk
Tierra home page: www.hip.atr.co.jp/~ray/tierra References
Virtual Alife Library: www.cs.brandeis.edu/~zippy/alife-library.html
1. C.G. Langton, Artificial Life: An Overview,
MIT Press, Cambridge, Mass., 1995, p. ix.

6 IEEE INTELLIGENT SYSTEMS

 2. C.G. Langton, Artificial Life: An Over- Robotics “Some rehabilitation centers around
view, MIT Press, Cambridge, Mass., the world are starting programs that will
1995, p. 113.
NASA robotics allow therapists to train individuals af-
fected with spinal injuries, stroke, and
3. K. Sims, “Evolving 3D Morphology and
Behavior by Competition,” Proc. Artificial could help patients perhaps other neuromotor disorders to
Life IV, R. Brooks and P. Maes, eds., MIT
Press, Cambridge, Mass., 1994, pp. walk improve their mobility and stepping ca-
28–39. pacity,” says Reggie Edgerton, professor
NASA engineers and neurophysiologists in the departments of physiological sci-
4. C. Adami, Introduction to Artificial Life, at the University of California, Los Angeles, ence and neurobiology at UCLA. “This
Springer-Verlag, New York, 1998, p. 13. are creating a robot-like device that could robotic device could help therapists in
5. H. Lipson and J. Polluck, “Automatic help rehabilitate people with spinal cord those rehabilitation efforts.”
Design and Manufacture of Robotic injuries. JPL robotic engineers have worked
Lifeforms,” Nature, Vol. 406, No. 6799, “We are developing a prototype ro- alongside therapists to develop the de-
2000, pp. 974–978. botic stepper device that, when com- vice, which has highly sensitive sensors
6. C. Adami, Introduction to Artificial Life, plete, will be used as part of rehabilita- that collect up to 24 different data read-
Springer-Verlag, New York, 1998, p. 15. tion that can potentially help some peo- ings of the patient’s activity. The device,
7. G. Theraulaz and W. Bonabeau,
ple now wheelchair-bound take their connected to a computer, displays the
“Coordination in Distributed Building,” first steps,” says Jim Weiss, program information on the screen for the thera-
Science, Vol. 269, No. 687, 1995, pp. manager for collaborative neural repair pist to monitor.
686–688. at NASA’s Jet Propulsion Laboratory. According to Weiss, this same device
8. J.L. Casti, Would-be Worlds: How “This system can do the work of four could also someday be useful to astronauts
Simulation Is Changing the Frontiers of therapists and help monitor a patient’s and help them walk safely after prolonged
Science, John Wiley & Sons, New York, progress in a controlled manner.” periods in space, such as extended mis-
1997, p. 165. The device will look like a treadmill sions on the International Space Station.
9. T. Appenzeller, “Test Tube Evolution with robotic arms and will be fitted with a The robotic stepper device is one of sev-
Catches Time in a Bottle,” Science, Vol. harness to support the patient’s weight. eral projects in the Neural Repair Program
284, No. 5423, 1999, pp. 2108–2110. The arms resemble knee braces that attach at the UCLA Brain Research Institute
10. D.E. Nilsson and S. Pelger, “A Pessimistic to the patient’s leg, guiding the legs prop- (www.medsch.ucla.edu/som/bri) and JPL
Estimate of The Time Required for an Eye erly on the moving treadmill. Although the (www.jpl.nasa.gov). JPL and UCLA are
to Evolve,” Proc. Royal Society London B, robotic stepper is still in development and actively pursuing efforts to commercialize
London, 1994, pp. 53–58. is not yet ready for use in rehabilitation, the robotic system. JPL technically sup-
11. S. Levy, Artificial Life, Vintage Books, the device could be part of clinical trials at ported UCLA in filing a patent application
New York, 1992, p. 139. UCLA in about three years. in August.