Page 192 - Biomimetics : Biologically Inspired Technologies

Bar-Cohen : Biomimetics: Biologically Inspired Technologies DK3163_c006 Final Proof page 178 21.9.2005 2:56am

178 Biomimetics: Biologically Inspired Technologies

in particular, especially robotics’ ambition to physically interact with the world usually guided by
bio-inspired intelligent control systems.
Though many robots bear little semblance to animals, in some regards, robotics’ relationship
to biology is essential. In their more complete, integrated incarnations, robots can be considered to
be synthetic animals containing the perceptual systems, intelligence, motility, and mobility needed
to operate in the world in ways that are derivative of animals or inspired by them. To wit, the
accelerating functionality of robots can be largely traced to the widening intersection of biology and
engineering.
The biosciences and robotics are each exhibiting unprecedented, explosive rates of discovery
and innovation (Kurzweil, 2002), and their junction increasingly operates as a hybrid discipline
known by several alternate names: biomimetics, biorobotics, and bio-inspired robotics being
among them. As bio-discoveries are translated into robotics, the resulting robots look, act, and
function ever more like animals — synthetic organisms which then offer biologists opportunities to
test theories of animal locomotion, intelligence, and materials. This synergistic interplay represents
a feedback loop that propels the pace of discovery and innovation in both biology and technology
and this interplay between biology and robotics is largely the subject of this chapter.
The concept is not new. For thousands of years, people have sought to emulate animals and
people in various media; and as far back as ancient Egypt and Greece, the media included
mechanical automata (Cassell, 2001). In the 19th and 20th centuries, advances in electromotors,
batteries, materials, and manufacturing enabled these artificial creatures to move with increasingly
lifelike grace and autonomy. While bodies of robots do not yet possess the full capabilities of
humans or animals, the accomplishments and the pace of progress in these areas exceed those of
any other time in history. Already, robots walk and run bipedally in the fashion of humans (Doi,
2004), affect realistic facial expressions (Hanson et al., 2003), fly like hummingbirds (Dickinson,
2001), and perform many other animal-like feats.
In parallel with the progress of synthetic bodies, artificial brains have evolved swiftly as well.
With the 20th century information sciences, technology began to emulate the nervous system,
beginning with the work of McCulloch and Pitts, Turing, Walter and others (McCorduck, 1979).
From these pioneers, the work has continued with new generations of robotics and artificial
intelligence (AI) researchers. In AI research, engineers coordinate with biosciences under the
rubric of cognitive science to the profound benefit of both robotics and neuroscience. Since the
foundations of AI research in the 1940s, AI systems have increased enormously in functionality,
and have become widely deployed in commercial applications, such that AI now constitutes a
nearly $9 billion market (BCC, 2003a,b).
While most pioneering AI research focused on a ‘‘top-down’’ symbolic approach that largely
disregarded the importance of a body or embodiment, the work of Brooks and others pushed the
paradigm of intelligence as forming from ‘‘bottom-up’’ — highly distributed and physically-
embodied architectures (Brooks, 1991). This perspective reinvigorated the emulation of animal
and human bodies in robotics, and is validated by much bio and neuroscience, including the work of
Damasio (1994) , which shows that the mind and body are integrally connected, that the abstract
mind does not float mysteriously above the organism as Descartes postulated. In addition to
validating research in locomotion, sensing and grasping, this paradigm shift frees researchers and
companies to consider biomimetic displays of anthropomorphic emotion, animated by software
models of ‘‘emotional intelligence’’; and together these tools are characteristic of the booming field
of social robotics (Breazeal, 2002).
Although rhetoric generally discusses terms of ‘‘top-down’’ or ‘‘bottom-up’’, most progress is
clearly being rendered in patches of the middle actually, somewhere between top and bottom. The
spreading complexity of this patchwork progress mirrors the ‘‘systems paradigm’’ of biosciences. In
this paradigm, an organism, even an intelligent human organism, is a highly integrated web of
systems. For example, it is well demonstrated that the intelligence of human brain suffers terribly
should any of several ‘‘lower’’ brain systems (such as the amygdala) be damaged (Damasio, 2005).