Abstract
Air
muscle is essentially a robotic actuator which is replacing the conventional
pneumatic cylinders at a rapid pace. Due to their low production costs and very
high power to weight ratio, as high as 400:1, the preference for Air Muscles is
increasing. Air Muscles find huge applications in biorobotics and development
of fully functional prosthetic limbs, having superior controlling as well as
functional capabilities compared with the current models. This paper discusses
Air Muscles in general, their construction, and principle of operation,
operational characteristics and applications.
Introduction
Robotic
actuators conventionally are pneumatic or hydraulic devices. They have many
inherent disadvantages like low operational flexibility, high safety
requirements, and high cost operational as well as constructional etc. The
search for an actuator which would satisfy all these requirements ended in Air
Muscles. They are easy to manufacture, low cost and can be integrated with
human operations without any large scale safety requirements. Further more they
offer extremely high power to weight ratio of about 400:1. As a comparison
electric motors only offer a power ration of 16:1. Air Muscles are also called
McKibben actuators named after the researcher who developed it.
History
It was in 1958 that R.H.Gaylord
invented a pneumatic actuator which’s original applications included a door
opening arrangement and an industrial hoist. Later in 1959 Joseph.L.McKibben
developed Air Muscles. The source of inspiration was the human muscle itself,
which would swell when a force has to be applied. They were developed for use
as an orthotic appliance for polio patients. Clinical trials were realisd in
1960s. These muscles were actually made from pure rubber latex, covered by a
double helical weave (braid) which would contract when expanded radially. This
could actually be considered as a biorobotic actuator as it operates almost
similar to a biological muscle.
Air Muscle Schematic- McKibben Model
The
current form air muscles were developed by the Bridgestone Company, famous for
its tires. The primary material was rubber i.e. the inner tube was made from
rubber. Hence these actuators were called ‘Rubbertuators’. These developments
took place around 1980s.
Later in 1990s Shadow Robotic
Company of the United
Kingdom began developing Air Muscles. These
are the most commonly used air muscles now and are associated with almost all
humanoid robotic applications which were developed recently. Apart from Shadow
another company called The Merlin Humaniform develops air muscles for the same
applications, although their design is somewhat different from the Shadow
muscles.
Construction
The
Air Muscle consists of an inner rubber tube, which is often made from pure
rubber latex. It is surrounded by a
braided mesh.
Air muscle
construction [Shadow air muscle: 30mm]
The header at each end of the muscle consists
of an Aluminium ring, and a
Delrin plastic bung, with a female thread.
This thread can be used as a means of attachment, and to allow air into or out
of the muscle. The muscle is supplied with two Delrin fittings also.
Working
The
inner rubber tube is inflated by entering air at a pressure, usually limited to
3.5 bar. The movement of this tube is constrained by the braid. When the tube
gets inflated it experiences a longitudinal contraction. This would create a
pull at both ends of the tube. Usually one end of the tube will be attached to
somewhere so that force can be applied from one end. This pull when effectively
utolised could provide the necessary motion. The working of the Air Muscle
closely resembles that of the natural muscle and hence the name Muscle given to
it along with Air. The figure below shows the physical appearance of the muscle
at different stages of its working.
Air
Muscle at different stages
Theoretical Model
Using conservation of energy and
assuming the actuator maintains dV dP equal
to zero, reasonable for actuators built with stiff braid fibers that are always
in contact with the inner bladder, the tensile force produced can be calculated
from:
Where,
P - the input actuation pressure,
dV - the change in the actuator’s
interior volume
dL -
the change in the actuator’s length
Vb -
the volume occupied by the bladder
dW -
the change in strain energy density(change in stored energy/unit
volume).
Ff
describes the lumped effects
of friction arising from sources such as contact between the braid and the
bladder and between the fibers of the braid itself. Neglecting the second and
third terms on the right hand side of above equation and assuming the actuator
maintains the form of a right circular cylinder with an infinitesimally thin
bladder yields known solutions. The solution to the second term on the right
side of the equation is based on a non-linear materials model developed by
Mooney and Rivlin in the 1940’s and 1950’s proposed a relationship between stress
(σ ) and strain (ε ) given by σ = dW dε
where W is the strain
energy density function. Using the
assumptions of initial isotropy and incompressibility, W can be described
as a function of two strain invariants ( I1 and I2
):
where Cij are
empirical constants. Only two Mooney-Rivlin constants (C10
=118.4 kPa
and C01 =105.7 kPa)
were necessary for accurate results with the natural latex rubber bladder,
however, other materials may require additional constants. For the case of the
McKibben actuator, the experimental methods required to determine these
constants are dramatically simplified because the McKibben actuator’s strain
invariants, constrained by braid kinematics, are nearly the same as the strain
invariants for uniaxial tension .This fortuitous relationship eliminates the
need for multi-axial testing that would otherwise be necessary. Solving
equation a using the non-linear Mooney-Rivlin materials model results in
a McKibben actuator model whose structure is allowed to deform as well as store
elastic energy in a non-linear fashion. This model is given by:
where Fmr is the
predicted force, and parameters N , Lo , B, and
Ro are shown in figure 1 and figure 2. Bladder thickness is
denoted by to and is used in the bladder volume calculation.
λ1 refers to the actuator’s longitudinal stretch ratio and is given
by
λ1 = Li/ Lo, where Li
is the actuator’s instantaneous length and Lo is the
original, resting state length.
Figure:
1
Figure:
2
McKibben actuators are fabricated from two principle
components: an inflatable inner bladder made of a rubber material and an
exterior braided shell wound in a double helix. At ambient pressure, the actuator
is at its resting length (figure: 1). As pressure increases, the actuator
contracts proportionally until it reaches its maximally contracted state at
maximum pressure (figure: 2). The amount of contraction is described by the
actuator’s longitudinal stretch ratio given by λ1 = Li Lo where L is
the actuator’s length, and subscript i refers to the instantaneous
dimension and the subscript o refers to the original, resting state
dimension.
Estimation of Frictional Effects
The third term on the right of
equation a represents these frictional losses which are a function of (1) braid material, (2) bladder material, (3) pressure, and (4) actuator length. In lieu of a model that incorporates a function
for each of these, we have taken the intermediate step of lumping all of these
effects into a single parameter ( Ff ) as a simple function
of pressure. Analysis of the experimental data and theory predictions ( Fmr
) suggests a linear form given by:
Ff
= mP + b
where m and b are
empirically determined constants. The actuator model, which now includes the
geometry of the braid and bladder, the material properties of the bladder, and
a term for frictional effects (all three terms of equation a) is given by:
F = Fmr − Ff
A comparison of this model versus
experimental results for the largest actuator (nominal braid diameter of 1-1/4
in.) is presented in figure 3.5. The figure shows a reasonably close fit for
each of the four activation pressures tested. Similar results were obtained for
the two smaller actuators (nominal braid diameters of ¾ and 1/2) but are not
shown.
Figure: 3
Dynamic Properties
To measure the force-velocity
properties of the McKibben actuator, a series of experiments were conducted with
the axial-torsional Bionix (MTS Systems Corp., Minnesota ,
U.S.A. )
tensile testing instrument. Actuators of three sizes were constructed and tested. Each experiment
measured the force output at a constant pressure over the contraction range at various
velocities. One end of the actuator was rigidly attached to the load cell while
the other end was moved in response to the instrument’s digital controller.
Step velocity profiles were applied such that one end of the actuator was
rapidly accelerated and held to a constant velocity until the end of the
actuator’s working length was reached. Input step velocity profiles tested
included 1, 10, 25, 50, 100, 150, 200, 250, and 300 mm/s for concentric
contractions and 1, 10, 25, 50, 100, and 150 mm/s for eccentric contractions. Up
to 500 mm/s is possible; however, instantaneous fluctuations in velocity of 15
percent were measured during trails at 500 mm/s. The magnitude of these
fluctuations decreased at lower velocities, and was less than 9 percent at 300
mm/s and 6 percent at 200 mm/sec. This anomaly is thought to arise from the hydraulic
pump.
Experimental
Results
The experimentally measured output
force of a single McKibben actuator, plotted as a function of both length and
velocity, is shown in figure 4. The results shown are from an actuator whose
nominal braid diameter was ¾ inch and constructed with a natural latex bladder.
The actuator pressure was 5 bar and the original, resting state length of the
actuator was 180 mm. The output force is clearly a function of length, but not
of velocity. Similar results were obtained at lower pressures and with the
other two sized actuators, but are not shown.
Figure:
4
Operating Characteristics
The characteristics of Muscles as given by the Shadow
Robotic company.
45
N load
Specifications for a typical Air Muscle (the Shadow company)
Diameter
|
Weight
|
Pull (3.5 bar)
|
Maximum pull
|
Length
|
30
mm
|
80
g
|
35
kg
|
70
kg
|
290
mm (stretched)
|
stretched
form
These
measurements are taken when the muscle is fully stretched out, under a load of
at least 50N, and a pressure of 0 bar.
Hole – Hole Spacing 290mm
Total Muscle Length 250mm
Active Length 230mm
1: The Hole-Hole spacing is the distance between the
holes in the fittings at either end of the muscle. This is adjustable, as the
fittings can be
screwed in or out. They can also be removed
entirely, creating a more compact muscle. Use an M10 screw instead, and
remember to use
PTFE tape to ensure a good seal.
2: The Total Muscle Length is the length of the
whole muscle, excluding the fittings.
3: The Active Length is the length of the part of
the muscle which contracts under pressure, and does not include the headers.
These measurements are taken when the
muscle is pressurised to 3bar, with a load of 50N.
Hole – Hole
Spacing 210mm
Total Muscle
Length 170mm
Active Length
150mm
Contracted
form
Differences from pneumatic cylinders
The Air Muscle is a low pressure
actuator with a set of operational features unique in the field of robotics and
automation
a) - Smooth jerk free motion from start to finish due to the complete lack of stiction, the feature of standard actuators which produces the characteristic jerk so well known in air operated devices.
b) - Compliance - Although they can produce the force needed to move a function the Air Muscle will also yield when an obstacle is encountered, thus preventing damage to the object and the "Arm" - a distinct advantage where robots mix with humans.
c) - Light weight - The materials from which the Air Muscle is made are non-metallic and give it a relatively high power-to-weight ratio - a critical feature in choosing an actuator for a mobile robot
a) - Smooth jerk free motion from start to finish due to the complete lack of stiction, the feature of standard actuators which produces the characteristic jerk so well known in air operated devices.
b) - Compliance - Although they can produce the force needed to move a function the Air Muscle will also yield when an obstacle is encountered, thus preventing damage to the object and the "Arm" - a distinct advantage where robots mix with humans.
c) - Light weight - The materials from which the Air Muscle is made are non-metallic and give it a relatively high power-to-weight ratio - a critical feature in choosing an actuator for a mobile robot
Advantages of Air Muscles
Power to weight ratios in excess of 1
kW/kg, by way of comparison, electric drives typically has some 100 W/kg
A varying force-displacement relation
at constant gas pressure, contrary to pneumatic cylinders, which results in a
muscle-like behavior; an adjustable compliance, due to gas compressibility and
the dropping force-displacement characteristics
A maximum displacement or stroke of
up to 50% of initial length
The absence of friction and
hysteresis, as opposed to other types of PAMs
The ability to operate at a wide
range of gas pressures, and thus to develop both very low and very high pulling
forces
The possibility of direct connection
to a robotic joint, i. e. without having to use any gears,
because of their high output forces at all speeds.
Some of the advantages spelt out by
the shadow company typical to their products are:-
Lightweight - Air Muscles weigh as little as 10
gm - particularly useful for weight-critical applications
Lower Cost - Air Muscles are cheaper to buy and
install than other actuators and pneumatic cylinders
Smooth - Air Muscles have no 'stiction' and
have an immediate response. This results in smooth and natural movement.
Flexible - Air Muscles can be operated when
twisted axially, bent round a corner, and need no precise aligning.
Powerful - Air Muscles produce an incredible
force especially when fully stretched.
Damped - Air Muscles are self-dampening
when contracting (speed of motion tends to zero), and their flexible material
makes them inherently cushioned when extending.
Compliant - Being a soft actuator, Air Muscles
systems are inherently compliant.
Efficient - a muscle length can be maintained
with minimal energy input.
Fast -full contraction can be achieved in
less than one second from rest.
Disadvantages
The force which can be applied is
only tensile in nature. For both kinds of forces additional mechanisms are
required.
The efficiency of Air Muscles is not
as good as electric motors
Its total displacement is only about
20% to 30% of its initial length
Friction between the netting and the
tube leads to a substantial hysteresis in the force-length characteristics; this
obviously has an adverse effect on actuator behavior and necessitates using
complex models and control algorithms
Rubber is often needed to avoid the
tube from bursting, this comes at the cost of a high threshold pressure—typically
about 90 kPa —that has to be overcome in order to start deforming the rubber
material and below which the actuator will simply not operate
Rubber deformation, like any material
deformation, needs energy, this will lower the force output of this type of
muscle up to 60%.
Applications
Humanoid robots
The
major application of Air Muscles is in the field of humanoid robots. As these
actuators nearly resemble the characteristics of actual skeletal muscles, they
can perform a verity of functions as is performed by the human hand. Coupled
with the implementation of neural networks and powerful, precise sensors they
are capable of high end applications such as assembling of very minute
components etc.
Humanoid robot manufactured
by Shadow robotic company
Artificial limbs
Artificial limb developed at the
bio robotics Lab, University
of Washington .
At the bio robotic lab of university of Washington the limb as shown figure was
developed. The major requirements of their research team were:
1. Continuous and extended operation for
about 8-10 hours.
2. Low weight
3. Quieter operation
4. User satisfaction
5. No maintenance or low levels of
maintenance.
To satisfy all the fore mentioned requirements
to be satisfied, a research team might spend years. But partially these feats
were accomplished. The figure given below illustrates this.
Merlin Humaniform Air
Muscles attached to human hand
The Dexterous hands
The dexterous hand was developed
by the Shadow robotic company. The hands operate just like human hands with
five fingers. It is powered by 28 Air Muscles. The size is almost same as human
hands as they closely fit into a human hand. The figure shown illustrates this
fact.
The muscle can perform any function
the human hand performs. Besides it is equipped to swivel its fingers. It makes
use of 28 Air muscles for these movements. The human hand has 24 muscles. The
additional four in case Dexterous hands due to the swiveling motion.
Further
developments
The
Pleated Pneumatic Air Muscles [PPAMs]: As a result growing research in the
field of Air Muscles, another variant called pleated pneumatic air muscles were
developed. Pleated pneumatic artificial muscles are strong and lightweight
actuators that perform very well in position control and other automation and robotic
tasks. They are easy to use, require no gearing and are easy to connect and replace.
A high degree of positioning accuracy is accomplished with them and this just by using
off-the-shelf pressure regulating servo-valves together with simple PI control
techniques. Furthermore, they can easily be made to have a soft touch so as not
to damage fragile objects or to effect a safe man-machine interaction. Because
of their inherent characteristics PPAMs are suitable for powering walking and
running machines. Autonomous machine operation can then be guaranteed in a
number of ways, e. g.by using on-board small size internal
combustion engines.
Some of the stated advantages of
PPAMs are:-
A maximum displacement or stroke of
up to 50% of
initial length;
The absence of friction and
hysteresis, as opposed to other types of PAM.
Conclusion
Even
though Air Muscles are not capable
of offering an extremely wide range of operations, but in the case of
artificial legs, humanoid robots etc they offer a wide range of possibilities.
With further developments in neural networks and sensor equipments, it might be
possible replace an entire limb for an amputee and function normally like a
natural limb would do. The only draw back lies in developing a complete
theoretical model for calculating the characteristics such as fatigue etc.
Research is also directed towards substituting for Air with nitrogen or other
gases for maximum efficiency and better damping.
References
Published Papers [source:
The internet]
Measurement and Modeling of
McKibben Pneumatic
Artificial Muscles
Ching-Ping Chou
Blake Hannaford
Department
of Electrical Engineering
FT-10
Pleated Pneumatic Artificial Muscles: Compliant Robotic Actuators
Frank DAERDEN, Dirk LEFEBER, Bj¨orn VERRELST, Ronald VAN HAM
VRIJE UNIVERSITEIT BRUSSEL
Department of Mechanical
Engineering / Multibody Mechanics Group
Pleinlaan 2, 1050 Brussel , Belgium
frank.daerden@vub.ac.be
Dynamic Pneumatic Actuator Model for a Model-Based Torque Controller
Joachim Schr¨oder_
Center for Intelligent
Systems
http://eecs.vanderbilt.edu/CIS/
joachim.schroeder@vanderbilt.edu
Duygun Erol
Center for Intelligent
Systems
http://eecs.vanderbilt.edu/CIS/
Kazuhiko Kawamura, Ph.D.
Center for Intelligent
Systems
http://eecs.vanderbilt.edu/CIS/
R¨udiger Dillmann, Dr.-Ing.
Industrial Applications of
Informatics and Microsystems
http://wwwiaim.ira.uka.de/
Design and Construction of
An Artificial Limb Driven by
Artificial Muscles for
Amputees
Sunton
Wongsiri
Department of Orthopaedic
Surgery
Prince of Songkla University ,
Hat Yai,
Phone: 66-7445-1601, Fax:
66-7421-2915
Email:
wsunton@medicine.psu.ac.th
Sathaporn
Laksanacharoen
Mechanical Engineering
Department,
King Mongkut’s Institute of Technology North Bangkok ,
Phone: 662-913-2500, Fax:
66-2586-9541
Email: STL@kmitnb.ac.th
Artificial Muscles: Actuators for Biorobotic Systems
Glenn Kenneth Klute
A dissertation submitted in
partial fulfillment of the
requirements for the degree
of
Doctor of Philosophy, University of Washington ,1999
Fatigue
Characteristics of McKibben Artificial Muscle Actuators
Glenn K. Klute
Department
of Bioengineering
gklute@u.washington.edu
Blake Hannaford
Department
of Electrical Engineering
blake@ee.washington.edu
http://rcs.ee.washington.edu/BRL/
Accounting for Elastic
Energy Storage in
McKibben Artificial
Muscle Actuators
Glenn K. Klute
Department of
Bioengineering
gklute@u.washington.edu
Blake Hannaford
Department of Electrical
Engineering
blake@ee.washington.edu
http://rcs.ee.washington.edu/BRL/
Books
Advanced Mechanics of Solids
L S Srinath
Tata McGraw-Hill Publishing
Company Limited, New Delhi .
A Text Book of Metallurgy and Material Science
O P Khanna
Dhanpat Rai Publications, New Delhi
Websites
www.shadow.org.uk
www.brl.washington.edu
www.robotstoreuk.com
www.kis.uiuc.edu
http://teams.kipr.org
http://biorobotics.cwru.edu
EmoticonEmoticon