Technical Paper Presentation Contest

ARTIFICIAL INTELLIGENCE
(Biologically Inspired Intelligent Robots Using Artificial Muscles)

Abstract: Why Artificial Intelligence?

Artificial Intelligence is a branch Motivation...
of Science which deals with helping Computers are fundamentally well
machines finds solutions to complex suited to performing mechanical
problems in a more human-like fashion. computations, using fixed programmed rules.
This generally involves borrowing This allows artificial machines to perform
characteristics from human intelligence, and simple monotonous tasks efficiently and
applying them as algorithms in a computer reliably, which humans are ill-suited to. For
friendly way. A more or less flexible or more complex problems, things get more
efficient approach can be taken depending difficult... Unlike humans, computers have
on the requirements established, which trouble understanding specific situations, and
influences how artificial the intelligent adapting to new situations. Artificial
behavior appears. Intelligence aims to improve machine
Humans throughout history have always behavior in tackling such complex tasks.
sought to mimic the appearance, mobility,
functionality, intelligent operation, and
thinking process of biological creatures. This When will Computers become truly
field of biologically inspired technology, intelligent?
having the moniker biometrics, has evolved Limitations...
from making static copies of human and
To date, all the traits of human
animals in the form of statues to the
intelligence have not been captured and
emergence of robots that operate with
applied together to spawn an intelligent
realistic appearance and behavior. This
artificial creature. Currently, Artificial
paper covers the current state-of-the-art and
Intelligence rather seems to focus on lucrative
challenges to making biomimetic robots
domain specific applications, which do not
using artificial muscles.
necessarily require the full extent of AI
capabilities. Researchers know this limit of
machine intelligence as narrow intelligence.
Keywords: EAP, artificial muscles, artificial
There is little doubt among the community
intelligence, biometrics
that artificial machines will be capable of
Introduction: intelligent thought in the near future. It's just a
AI is generally associated with question of what and when. The machines
Computer Science, but it has many important may be pure silicon, quantum computers or
links with other fields such as Math’s, hybrid combinations of manufactured
Psychology, Cognition, Biology and components and neural tissue. As for the date,
Philosophy, among many others. Our ability expect great things to happen within this
to combine knowledge from all these fields century!
will ultimately benefit our progress in the
quest of creating an intelligent artificial being.
How does AI work ? speech, and perform mobility very similar to
Technology... living creatures including walking, hopping,
and swimming. Further, advances in polymer
There are many different sciences led to the emergence of artificial
approaches to Artificial Intelligence, none of muscles using Electro active Polymer (EAP)
which are either completely right or wrong. materials that show functional characteristics
Some are obviously more suited than others in remarkably similar to biological muscles.
some cases, but any working alternative can Making creatures that behave like the
be defended. Over the years, trends have biological model is a standard procedure for
emerged based on the state of mind of the animatronics industry that is quite well
influential researchers, funding opportunities graphically animates the appearance and
as well as available computer hardware. behavior of such creatures. However,
Over the past five decades, AI engineering such biomimetic intelligent
research has mostly been focusing on solving creatures as realistic robots is still challenge
specific problems. Numerous solutions have due to the need to physical and technological
been devised and improved to do so efficiently constraints.
and reliably. This explains why the field of
Artificial Intelligence is split into many Artificial muscles:
branches, ranging from Pattern Recognition to
Artificial Life, including Evolutionary Muscles are the key to the mobility
Computation and Planning. and manipulation capability of biological
creatures and when creating biomimetic it is
essential to create actuators that emulate
Artificial life through robotics: muscles. The potential to make such actuators
Laws of Robotics: is increasingly becoming feasible with the
1. A robot may not injure a human being or, emergence of the electro active polymers
through inaction, allow a human being to (EAP), which are also known as artificial
come to harm. muscles [Bar-Cohen, 2001]. These materials
2. A robot must obey the orders given it by have functional similarities to biological
human beings except where such orders would muscles, including resilience, damage
conflict with the first law. tolerance, and large actuation strains.
3. A robot must protect its own existence as Moreover, these materials can be used to
long as such protection does not conflict with make mechanical devices with no traditional
the first or second law. components like gears, and bearings, which
Robotics has been an are responsible to their high costs, weight and
evolution of the field of automation where premature failures. The large displacement
there was a desire to emulate biologically that can be obtained with EAP using low
inspired characteristics of manipulation and mass, low power and, in some of these
mobility. In recent years, significant advances materials also low voltage, makes them
have been made in robotics, artificial attractive actuators. The capability of EAPs to
intelligence and others fields allowing to make emulate muscles offers robotic capabilities
sophisticate biologically inspired robots [Bar- that have been in the realm of science fiction
Cohen and Brea zeal]. Using these advances, when relying on existing actuators.
scientists and engineers are increasingly
reverse engineering many animals'
performance characteristics. Biologically
inspired robotics is a subset of the
interdisciplinary field of biomimetics.
Technology progress resulted in machines that
can recognize facial expressions, understand
 What are they going to be used for? What
responsibilities will they carry? These are just
a few of the questions, which are on the minds
of many who ever stopped to think what AI is
about.
Robots, which could build other
robots, tried to protect humans from
everything until people could not do anything
by themselves. They became totally dependent
on the machines. The creator of these robots
could not destroy them because they would
keep him away from doing so. After all how
could they protect people after they would be
destroyed? How much responsibility and
FIGURE 1: A graphic illustration of the grand authority should these machines have? This
challenge for the development of EAP question becomes even more important when
actuated robotics – an arm wrestling match one considers a fact that one of the biggest
against human. supporters of AI research is military. One of
Unfortunately, the EAP materials that the reasons is that in "...a nuclear age, ... a new
have been developed so far are still exhibiting generation of very intelligent computers
low conversion efficiency, are not robust, and incorporating AI could actually defend the
there are no standard commercial materials country better, faster, and more rationally than
available for consideration in practical humans."4 Even if this is true we do not think
applications. In order to be able to take these that a machine should handle the
materials from the development phase to responsibility for the fate of human
application as effective actuators, there is a civilization, no matter how intelligent it is.
need for an established infrastructure. For this
purpose, it is necessary to develop
comprehensive understanding of EAP Biometric robots using EAP :
materials' behavior, as well as effective
processing, shaping and characterization Mimicking nature would
techniques. The technology of artificial significantly expand the functionality of
muscles is still in its emerging stages but the robots allowing performance of tasks that are
increased resources, the growing number of currently impossible. As technology evolves,
investigators conducting research related to great number of biologically inspired robots
EAP, and the improved collaboration among actuated by EAP materials emulating
developers, users, and sponsors are leading to biological creatures is expected to emerge.
a rapid progress. The Responsibility of AI The challenges to making such robots can be
Research and seen graphically in Figure 2 where humanlike
and dog-like robots are shown to hop and
express joy. Both tasks are easy for humans
Development The question that concerns and dogs to do but are extremely complex to
many in regards to the development of perform by existing robots.
machines capable of intelligence comparable
to that of a human being, is that of FIGURE 2: Making a joyfully hopping
consequences of creating servants capable of human-like and dog-like robots actuated by
becoming masters. This is a question in itself EAP materials are great challenges for
containing many more: How much power will biomimetic robots
these machines - (robots, as many like to think
of them) have? Who is going to control them?
 of making such a capability with high
resolution and small workspace was enabled
with the novel MEMICA system (Mechanical
Mirroring using Controlled stiffness and
Actuators) For this purpose, scientist at JPL
and Rutgers University used an EAP liquid,
called Electro-Rheological Fluid (ERF), which
becomes viscous under electro-activation.
Taking advantage of this property, they
designed miniature Electrically Controlled
Stiffness (ECS) elements and actuators. Using
this system, the feeling of the stiffness and
forces applied at remote or virtual
environments will be reflected to the users via
proportional changes in ERF viscosity.

Biologically inspired robots:

FIGURE 3: An android head and a robotic
hand that are serving as biomimetic platforms The evolution in capabilities that
for the development of artificial muscles. are inspired by biology has increased to a
level where more sophisticated and
demanding fields, such as space science, are
considering the use of such robots. At JPL, a
six-legged robot is currently being developed
for consideration in future missions to such
planets as Mars. Such robots include the
LEMUR (Limbed Excursion Mobile Utility
Robot). This type of robot would potentially
perform mobility in complex terrains, sample
acquisition and analysis, and many other
functions that are attributed to legged animals
including grasping and object manipulation.
This evolution may potentially lead to the use
of life-like robots in future NASA missions
that involve landing on various planets
Remote presence via haptic interfaces: including Mars.

Remotely operated robots and

simulators that involve virtual reality and the The details of such future missions will be
ability to “feel” remote or virtual environment designed as a plot, commonly used in
are highly attractive and offer unmatched entertainment shows rather than conventional
capabilities [Chapter 4 in Bar-Cohen and mission plans of a rover moving in a terrain
Brea zeal, 2003]. To address this need, the and performing simple autonomous tasks.
engineering community are developing haptic Equipped with multifunctional tools and
(tactile and force) feedback systems that are multiple cameras, the LEMUR robots are
allowing users to immerse themselves in the intended to inspect and maintain installations
display medium while being connected thru beyond humanity's easy reach in space with
haptic and tactile interfaces to allow them to the ability to operate in harsh planetary
perform telepresence and “feel" at the level of environments that are hazardous to human.
their fingers and toes. Recently, the potential This spider looking robot has 6 legs, each of
which has interchangeable end-effectors to by Breazeal [2002]. Kismet perceives a
perform the required mission (see Figure 4). variety of natural social cues from visual and
The axis symmetric layout is a lot like a auditory channels, and delivers social signals
starfish or octopus, and it has a panning to people through gaze direction, facial
camera system that allows omni-directional expression, body posture, and vocalizations.
movement and manipulation operations. Natural language processing will
provide important services for people who
speak different languages. If a computer is
FIGURE 4: A new class of multi-limbed able to understand natural languages, it will
robots called LEMUR (Limbed Excursion also be able to translate from one language to
Mobile Utility Robot) is under development at another. The "universal translator" widely
JPL used Star Trek may actually become a reality!
This, of course, also includes voice
recognition, or speech recognition.
. One of the big problems of humankind is that
people start thinking only after they get into
trouble. People like to invent new things but
most of the time they never stop to think about
consequences of their inventions. We think
that now is the time to raise the question about
Artificial Intelligence: "Should we do it?". As
we have already mentioned, there are a lot of
ethical and moral problems which could arise
with the future development of Artificial
Intelligence. Even though there are a lot of
advantages of using Artificial Intelligence in
the future, there are still big obstacles which
should make us hesitant to develop or use
robots in our lives.
Some may argue that this does not
have to be true and all people will be happy
and have a lot of free time to spend with their
Robots as part of the human society : families and for leisure. But others will quote
As robots are getting the Marx and say: "Labor is what makes a man
appearance and functionalities of humans and human." These people will argue that people
animals there is a growing need to make them will become dependent on AI, and they will
interact and communicate as a sociable partner loose the ability to provide for themselves, or
rather than a tool. This trend is requiring that even the ability to reason. Then humanity
robots would be able to communicate, would be reduced to the primordial animal
cooperate, and learn from people in familiar state.
human-oriented terms. Such a capability poses
new challenges and motivates new domestic, Future of Artificial Intelligence: should we
entertainment, educational, and health related do it?
applications for robots that play a part in our What kind of developments should
daily lives. It requires obeying a wide range of we expect in the area of Artificial
social rules and learned behaviors that guide Intelligence? Judging from the research topics
the interactions with, and attitudes toward, that we have today, we might predict that in
interactive technologies. Such robots are the near future things such as object
increasingly emerging and one example of recognition, voice recognition, and natural
such a robot is the Kismet that was developed language understanding will be a reality. Will
there be systems so advanced that they have to for computer games and robotic pets. Lets also
be given rights similar to those of humans? not forget big establishments dealing with
Probably not in the foreseeable future. But huge amounts of information such as
maybe in a little more than half a century, if hospitals, banks and insurances, who can use
the humanity survives that long, such AI to predict customer behavior and detect
machines may very well develop. trends.

What kind of advantages would future

Artificial Intelligence systems offer? Disadvantages:
1. Potential for malevolent programs,
They will probably be increasingly used “cold war” between two countries,
in the field of medicine. A knowledge based unforeseen impacts because it is
expert system, which can cross-reference complextechnology, environmental
symptoms and diseases will greatly improve consequences will most likely be
the accuracy of diagnostics. Object minimal.
recognition will also be a great aid to doctors. 2. Self-modifying, when combined
Along with images from cats cans or X-ray with self-replicating, can lead to
machines, they will be able to get preliminary dangerous, unexpected results,
analysis of those images. This of course will such as a new and frequently
be possible only if people solve legal mutating computer virus.
questions that arise by giving power to a 3. As computers get faster and more
machine to control or influence the health of a numerous, the possibility of
human. randomly creating an artificial
The most difficult question arises intelligence becomes real.
when we start to think about long term goals. 4. Military robots may make it
Do we want to build a computer which will be possible for a country to
like us? If we do, what do we need them for? indiscriminately attack less-
What will these human-computers do for advanced countries with few, if
humanity? Nobody has an answer to these any, human casualties.
questions and we can only speculate about the 5. Rapid advances in AI could mean
consequences. What is even more important, massive structural unemployment
scientist will not stop to try to achieve more 6. AI utilizing non-transparent
and more every year. We will just have to live learning (i.e. neural networks) is
and see what the future of Artificial never completely predictable.
Intelligence and the future of Human
existence, will be like.
• Idea of Artificial Intelligence is being Conclusion : (what should happen)
replaced by Artificial life, or anything • When programs that appear to
with a form or body. demonstrate sentience appear
• The consensus among scientists is that (intelligence and awareness), a panel
a requirement for life is that it has an of scientists could be assembled to
embodiment in some physical form, determine if a particular program is
but this will change. Programs may not sentient or not.
fit this requirement for life yet. • If sentient, it will be given rights, so, in
general, companies will try to avoid
Applications... developing sentient AI since they
The potential applications of Artificial would not be able to indiscriminately
Intelligence are abundant. They stretch from exploit it.
the military for autonomous control and target • Software companies should be made
identification, to the entertainment industry legally responsible for failings of
 software that result in damage to third 0-8194-4872-9 (2003), pp. 1-410
parties despite good-faith attempts at http://www.spie.org/web/abstracts/oepre
ss/PM122.html
control by the user.
• AI and robotics have the potentially to
truly revolutionize the economy by
replacing labor with capital, allowing
greater production—it deserves a
corresponding share of research
funding !

References:

Bar-Cohen Y. (Ed.), “Electroactive Polymer

(EAP) Actuators as Artificial Muscles -
Reality, Potential and Challenges,” ISBN
0-8194-4054-X, SPIE Press, Vol. PM98,
(March 2001), pp. 1-671
http://ndeaa.jpl.nasa.gov/nasande/
yosi/yosi-books.htm
Bar-Cohen Y. and C. Breazeal (Eds.),
“Biologically-Inspired Intelligent
Robots,” SPIE Press, Vol. PM122, ISBN