The
Problem:
Although artificial
intelligence has been very successful and is present, largely unseen,
in the every day life of most people in the developed world, there still
seems to be some things missing. Over the last fifteen years we have
built behavior-based artificial creatures situated in the world like
insects [2] and recently we have built robots with
which we can have human-like social interaction [1].
But we never quite forget that these systems are machines and not alive.
We build these models to better understand the biological systems, but
the models never work as well as biology. On the other hand scientists
have gotten very good at modeling fluids, materials, planetary dynamics,
nuclear explosions, and all manner of physical systems. We can put some
parameters into a program, let it crank, and get accurate predictions
of the physical character of the modeled system. But, we are not good
at modeling living systems, neither in the small, nor in the large.
Motivation:
Perhaps we are missing
something fundamental and currently unimagined in each of our various
models of behavior, perception, cognition, evolution, natural selection,
morphogenesis, etc. If this turns out to be true, then we will need
to have some new ways of thinking about the issues of living systems
if we are to make progress. This would be disruptive to all the sciences
of living systems. As an analogy, suppose we were building physical
simulations of elastic objects falling and colliding. If we did not
quite understand physics we might unfortunately leave out mass as a
specificable attribute of the objects. Their falling behavior would
at first seem correct, but as soon as we started to look at collisions
we would notice that the physical world was not being modeled correctly.
What might the nature
of this missing element be? One future possibility is that we will discover
some aspect of living systems that is invisible to us right now. The
current scientific view of living systems is that they are machines
whose components are biomolecules. One can not rule out that we will
discover some new properties of biomolecules or some new ingredient
that we do not currently understand, that are necessary for the aspects
of life which we observe. One might imagine that there is something
on a par with the discovery a century ago of radiation, which ultimately
led to our still evolving understanding of quantum mechanics. This is
now a firmly ingrained part of physics that was unimagined one hundred
years ago lots of good physics was done before its discovery.
Note that even now quantum mechanics is an arena that is trying to be
actively tamed in the service of fundamentally new classes of computation.
Of course, relativity was the other great such discovery of the twentieth
century with a similar disruptive impact on our basic understanding
of physics. Some such discovery might rock our understanding of the
basis of living systems, and indeed there could be different such discoveries
for various aspects of living systems, such as evolution, perception,
etc.
Or it may be that
what is missing is simply some "new mathematics". It may be
that we do not require any new physics to be present in living systems.
Rather it might turn out to be the case that we are simply not seeing
some fundamental mathematical description of what is going on in living
systems and we are leaving it out of all our AI and Alife models for
that reason. This argument is played out in more detail in [3].
This project is
an attempt to find this new mathematics.
Previous
Work:
There is of course
a long history of trying to understand what it is that separates living
from non-living matter. An insightful analysis of the nature of what
an adequate theory might be like can be found in [4].
Approach:
We have adopted
a three pronged attack on the problem.
1. We are
building robots and trying to equip them with capabilities that living
systems have, but which robots have not previously had. These include
self-repair, self-reproduction, energy self-sufficiency, growing bodies,
and control adaptation to variable body morphology. In this work we
demand a minimal competence from the robots, but that competence is
not the primary goal the primary goal is the exploration of the
capabilities that all living systems have, but which robots have not
previously had.
2. We are
building large scale computational experiments to explore aspects of
living systems that can not at this time be directly approached in hardware.
These include self-organization of pre-biotic chemistry, self-organization
of very simple neural systems in very primitive creatures, self-organization
of physical structures based on tensegrity, and the evolution of physical
attributes of creatures in complex environments.
3. We are
trying to generalize our results in both the first two areas to be able
to state mathematical theorems governing aspects of living systems.
This work lags well behind the first two areas.
None of these approaches
is completely unique. However, it is unique to have all three activities
going on in a single research group. We are hopeful that the three activities
will cross fertilize each other and lead to new insights.
Impact:
This work has two
potential sorts of impact. First, it may give better insight into how
living systems work, so that they can be better analyzed and understood.
Second, it may allow us to build new classes of machines with many of
the desirable properties of living systems, including their robustness,
their adaptability, and the minimal overhead that is required to fabricate
them.
Future
Work:
This project is
in its initial stages. There will be many changes to the specific problems
that we work on as we refine the questions we are asking. There are
however some interesting aspects of the problem that we have not yet
addressed. For instance, we are still using conventional materials for
our robots. In the future we intend to explore the use of different
materials. We may also consider the explicit use of bio materials.
References:
[1]
Cynthia Breazeal. Designing
Sociable Robots Cambrige, Massachusetts, 2001.
[2]
Rodney Brooks. Cambrian
Intelligence Cambridge, Massachusets, 1999.
[3]
Rodney Brooks. The
relationship between matter and life Nature, 409:409-411, 2001.
[4]
Robert Rosen. Life
Itself: A Comprehensive Inquiry Into the Nature, Origin, and Fabrication
of Life Columbia University Press, New York, 1991.
[back
to top]
|