today. We see present-day humanoid
robots such as Honda’s Asimo with a
huge backpack full of state-of-the-art
batteries and are a bit disappointed
that it can operate for less than an
hour. It makes you wonder just what
will be required in the way of a
battery or other type of system to
provide power for an exoskeleton that
must deliver much more force than
small experimental robots.
Most robotic exoskeletons have
utilized electrical systems and electric
motors to power the joints. Torque
requirements at the knee, elbow, and
shoulder joints are quite high and
many designers have used worm drive
gearing at these points. The placement
of large gearmotors at these locations
is difficult and some designers have
positioned the large motors at different
locations and have used flexible shafts
to deliver the rotational power to gear
assemblies at the joints.
All in all, the sum total of the
motors, gear systems, structural members, and batteries adds up to a lot of
mass and weight that must be moved
in addition to any prospective payloads.
As previously mentioned,
Monty Reed and others have used
compressed air systems for their suits.
This is due to the large amount of
force that a pneumatic cylinder can
exert, but even Monty realizes that
his future suits must contain a more
compact power system.
Pneumatic cylinders act just like
our muscles to pull or push an
appendage. In a human, the ulna
bone of the lower arm is pulled
upwards to the humerus bone of the
upper arm by the bicep muscle which
is attached by tendons. Think about
the tremendous force that muscle is
exerting on the bones when a person
is ‘curling’ a 100 pound barbell, for
example. At 50 pounds per arm of
barbell weight, and an assumed ratio
of 10:1 (arm length, tendon attachment point radius), each bicep must
pull at a force of 500 pounds each. A
pneumatic exoskeleton cylinder must
also be able to exert the same force.
Another design issue with pneumatic
and hydraulic exoskeletons is the
difficulty of varying the cylinder’s
pressure to accommodate differences in
load conditions. As you can imagine, a
barbell will exert the greatest load on
the arm’s cylinder when it is extended
at 90 degrees; proportionally less
pressure is required when the weight
is at a lower or higher position.
It is hard to rapidly change
pneumatic or hydraulic pressure using
a simple open/close solenoid valve. A
motorized valve will deliver the same
system high pressure when barely
cracked open; the flow rate and movement will just be slower, not the force.
Large robots have large cavities
for motors, actuators, and battery
systems. An exoskeleton must contain
a full-sized human being, as well as
those motors, actuators, power
sources, and control systems. The
human being must be able to enter
the suit easily, be comfortable, and
not claustrophobic, and have padding
and physical protection from injuries.
Control systems must be easily
reached and manipulated by
These caveats sometimes doom
an otherwise excellent design. If the
human takes up most of the inside
space, where do the batteries, joints,
motors, and controls go?
Another issue that I mentioned
earlier is the balancing problem.
Bipedal robots have sophisticated
algorithms to coordinate pace and
length of steps all the while controlling
momentum, acceleration and
deceleration, turning, balance, and
the positions of many joints. Inverted
pendulum and accelerometer
feedback systems can stabilize a robot
that normally walks at a specific pace
but a human wearing an exoskeleton
walking at various paces presents a
more critical balance problem.
Full robotic exoskeletons, versus
a system that augments only the
lower legs pose a mass/momentum
problem. Abrupt acceleration or
deceleration mass can cause the suit
to tip forward or backward.
Moving a large mass out in front
of the suit can also cause a tipping
problem. Designers compensate for
this unbalanced condition by the use
of pressure strain gauges at the front
and back of the feet to determine
a neutral balance point, and then
shifting the leg joints to compensate.
In the mid ‘60s, General Electric
engineers in their Schenectady, NY
facility came up with Hardiman,
shown in Figure 2. ‘Hardi’ is short
for Human Augmentation Research
and Development Investigation.
This massive, strength-amplifying
exoskeleton weighed 1,500 pounds
and touted a 750 pound lifting
capacity. Bomb loading, nuclear plant
operations, underwater construction,
and even space uses were envisioned
by GE. Hardi had 30 joints and two
massive arms interconnected by
hydraulic and electrical lines.
FIGURE 2. GE’s Hardiman 2 Exoskeleton.
SERVO 11.2009 77