Biomedical Engineering Reference
In-Depth Information
injected into the muscle of the residual limb, send signals wirelessly to a surface receiver
to control the prosthesis. Because they can be placed within individual muscle bundles,
the signals obtained offer much improved discrimination (Weir, Troyk et al., 2009).
10.9.4.7 Feedback
One of the problems with using myoelectric signals in isolation is that there is no pro-
prioception or tactile feedback. This requires that the amputee look at the prosthesis to
gauge position and force. One alternative is to use auditory feedback (Smith, 1990) or a
vibrotactile or electrotactile element (known as a tactor) secured against the user's skin to
convey some information about the angle of a joint or the force applied by the terminal
device. A good example is the pressure sensor on the Luke hand that generates a signal
proportional to the grip strength. This signal controls a tactor that vibrates slightly when
the grip is light, and as the user's grip tightens the frequency of the vibration increases.
This enables a user to pick up and drink out of a flimsy paper cup without crushing it or
to firmly hold a heavy cordless drill without dropping it. System dynamics should be con-
sidered when stimulating skin sensors. A response time of a few hundred milliseconds is
typical from the application of the stimulus to the user registering the effect. Additionally,
the skin becomes desensitized to continuously applied stimuli after a few minutes.
For grasping, it is impractical to provide feedback from each finger, so prosthetic
hands like the Gufu III include velocity control of each joint until contact, as described by
E
i
=
K
Vi
(θ
i
−
θ
di
)
(10.13)
θ
where
E
i
is the motor input,
K
Vi
is the velocity feedback gain, and
di
is the required
rotation rate of the
i
-th joint.
After contact, the grasping force on each link controls the rotation rate of the adjacent
joint,
K
Ii
θ
di
=−
K
Pi
(
F
i
−
F
di
)
−
(
F
i
−
F
di
)
dt
(10.14)
where
K
Pi
is the proportional force feedback gain,
K
Ii
is the integral force feedback gain,
and
F
di
is the required force of the
i
-th link.
This control strategy ensures that the object is grasped quickly and that the grasping
force is distributed reasonably uniformly across the links of each of the fingers.
A recent development is to use reinnervation of the pectoral segment to provide sensory
information directly to the nerves that used to carry the sensory signals from the hands.
However, though this technique can provide some tactile feedback it is not yet sufficiently
advanced to provide the required proprioception for reliable control of a prosthesis.
10.9.5 Leg Mechanisms
A prosthesis that combines intelligence and motorized actuation has the ability to regen-
erate the correct gait kinematics. This process involves programming the knee to execute
normal gait dynamics during all phases of the cycle. It must also be capable of meet-
ing other activities for which passive prosthetics are ill suited. These include climbing
and descending stairs or simply sitting down or standing up (Pons, 2008). Advances
in microprocessor speed, available memory, and low-power operation in conjunction
with developments in artificial intelligence are now making it more feasible to produce
