Page 273 - Biomimetics : Biologically Inspired Technologies

Bar-Cohen : Biomimetics: Biologically Inspired Technologies DK3163_c009 Final Proof page 259 21.9.2005 3:10am

Engineered Muscle Actuators 259

Figure 9.2 (See color insert following page 302) Left: axonal sprouting (A) from an explanted motor neuron
cell cluster (V) toward a target tissue (T), in this case, an aneural cultured skeletal muscle ‘‘myooid.’’ Right: a simple
cell culture system demonstrating axonal sprouting between neural (PC–12) and myogenic (C2C12) cell lines. This
co-culture system allows the study of synaptogenesis in culture. (Photographs taken by members of the Functional
Tissue Engineering Laboratory at the University of Michigan: Calderon, Dow, Borschel, Dennis.)

9.10 MUSCLE BIOREACTOR DESIGN FOR THE IDENTIFICATION,
CONTROL, AND MAINTENANCE OF MUSCLE TISSUE

The engineering of complex functional tissues such as skeletal muscle is by definition a systems
engineering problem. Functional muscles are composed of a number of highly integrated tissue
systems, none of which is known to function in isolation for any significant period of time without
massive deterioration in performance. Any attempt to engineer a functional muscle tissue system
ex vivo, and to employ that muscle system as a source of motility in robots or prostheses, will by
necessity require the development of bioreactor technologies to (1) guide the tissue development to
the desired phenotype ex vivo, (2) maintain the tissue at the desired phenotype while it is performing
its function, and (3) control the mechanical output of the tissue through electrical stimulation.
Critical to these three objectives are bioreactor technologies that are capable of monitoring and
controlling a muscle’s mechanical and electrical environment.
In Figure 9.3, a muscle bioreactor is shown that can implement muscle identification, control,
and maintenance protocols under generalized boundary conditions while also providing flexible
feedback control of electrical stimulation parameters (Farahat and Herr, 2005). These features are
accomplished by having two real-time control loops running in parallel. The first loop, or the
mechanical boundary condition (MBC) control loop, ensures that the mechanical response of
the servo simulates the dynamics of the associated muscle boundary condition. For example,
if the desired boundary condition is a second order, mass–spring–damper system, the MBC control
loop controls the motion of the end points of the muscle–tendon unit as if the muscle–tendon were
actually pulling against physical mass–spring–damper mechanical elements. The MBC control
loop allows for a whole host of boundary conditions, from finite (but nonzero) to infinite impedance
conditions. Clearly, to understand muscle tissue performance, muscle dynamics, and the dynamics
of the load for which the muscle acts upon must be taken into consideration. Examples of finite-
impedance boundary conditions include loads such as springs, dampers, masses, viscous friction,
coulomb friction, or a combination thereof. Such loads prescribe boundary conditions that are
generally defined in terms of dynamic relationships between force and displacement. Under these
loading conditions, it would be expected that the dynamics of the load will interact with the
contraction dynamics of the muscle, leading to a behavior that is a resultant of both. This is