13th Bristol International RPV Conference,

Bristol England, 30 March - 1 April 1998

UPDATE ON FLAPPING WING MICRO AIR VEHICLE RESEARCH
Ongoing work to Develop a Flapping Wing, Crawling Entomopter
Robert C. Michelson
Principal Research Engineer, Georgia Tech Research Institute
Adjunct Associate Professor, School of Aerospace Engineering, Georgia Institute of Technology
Past President, Association for Unmanned Vehicle Systems, International
Steven Reece
Graduate Student, School of Mechanical Engineering, Georgia Institute of Technology
ABSTRACT: An electromechanical multimode (fying/crawling) insect is being developed by
Robert Michelson and his design team at the Georgia Tech Research Institute. The project has received
initial IRAD funding from the Georgia Institute of Technology. The mechanical insect, known as an
Entomopter is based around a new development called a Reciprocating Chemical Muscle (RCM)
which is capable of generating autonomic wing beating from a chemical energy source. Through
direct conversion, the RCM also provides small amounts of electricity for onboard systems and fur-
ther provides differential lift enhancement on the wings to achieve roll and hence, steered fight. A
testbed for the RCM technique has been constructed and demonstrated. Trimmed stable short range
fight in a micro version of the entomopter is expected during 1998. In contrast to the testbed, entirely
different mechanisms will be used to implement the RCM in the fying version. This paper details
progress to date on the Entomopter development.
AUTHOR BIOGRAPHICAL SKETCH
Robert Michelson is the Technical Area Manager, Battlefeld Robotics & Unmanned Systems at the Georgia Tech
Research Institute. He is currently Director of the Department of Transportations Traffc Surveillance Drone project
as well as Director and Principal Investigator for the top-rated IRAD program at the Institute for FY97 involving the
development of a multimode mechanical insect-based micro air vehicle.
He has worked on and directed a number of remote sensing projects related to RF and radar applications. In ad-
dition he has performed verifcation and validation analyses of man-in-the-loop virtual reality fight simulators for
STRICOM. He directed a project to develop the avionics suite for an Air Force Robotic Air-to-Air Combat vehicle.
He directed a project to specify dual-mode IR/MMW seeker parameters for a lethal unmanned aerial vehicle (UAV)
system. He was responsible for generating remote fight control system specifcations for Soviet HAVOC and
HOKUM gunship drone simulators for the U.S. Army, and directed a task to develop a rotary winged UAV digital
stability augmentation system for MICOM.
As adjunct Associate Professor to the School of Aerospace Engineering, he teaches classes in avionics for UAVs. He
is also the creator and organizer of the annual International Aerial Robotics Competitions. Prior to joining the GTRI
staff, he participated in design and endo-atmospheric fight testing of computer-controlled space-based radar ocean
surveillance systems while employed by the Naval Research Laboratory in Washington, DC. He is author/coauthor
of more than 60 major technical publications.
Steven Reece has been instrumental in implementing the milli-scaled reciprocating chemical muscle testbed, testing
it, and compiling the performance data contained in this paper.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
N
othing in creation exhibits fxed wing fight be-
havior or propeller-driven thrust. Everything
that maintains sustained flight, uses flapping
wings. Even though there has been considerable analysis
in the literature of mechanisms for bird fight (Ellington
1
,
1984) and insect fight (e.g., Azuma
2
, 1992, and Brodsky
3
,
1994), and ornithopter-based (bird fight) machines have
been demonstrated nothing at the size level of an Ento-
mopter (entomo as in entomology + pteron meaning wing,
or a winged insect machine) has been tried.
An mechanical entomopter, or robotic insect, capable of
short distance trimmed fight and ground locomotion using
a reciprocating chemical muscle technique is currently
under development by the authors. In addition, autono-
mous navigation schemes (homing) based on Georgia
Techs integrated-optic interferometric waveguide sensor
will be developed as a means of directing controlled fights
and crawling in the future
Major Hurdles
Beyond the challenges of low Reynolds number aero-
dynamics (inertial force of body viscous force of air),
three major system-specifc technological areas must be
addressed before a any practical MAV can be felded.
These are:
NONSCALING ITEMS
STORED ENERGY
PROPULSION
NONSCALING ITEMS may be functions of external fac-
tors such as established GPS frequencies over which
there is little control. For example antennas may be of
suboptimal gain or directivity in order to ft the form fac-
tor of a MAV, while ground station frequencies may of
necessity, preclude anything but line-of-sight operation. A
reconnaissance MAV operating line-of-sight at a distance
of several kilometers may require an operational altitude
of several thousand feet in order to clear tree lines, hills,
and cultural items. The cost of being small becomes of
questionable beneft when the mission envelope begins to
overlap that of existing assets which can perform the same
reconnaissance mission.
STORED ENERGY becomes a significant impediment as
MAV mission duration increases. The present state-of-
the-art in battery technology does not allow for long en-
durance MAV missions, though it is hoped that someday
improved electrical storage media (carbon-air, fuel cells,
etc.) will result in the energy densities required for useful
long endurance (> 1 hour) missions in MAV-sized vehicles.
Near term solutions to onboard energy storage will come
from chemical or fossil fuels because of their superior
energy density. As a point of comparison, consider the
amount of releasable energy stored in a drop of gasoline
compared to that which can be stored in a battery the size
of a drop of gasoline.
Given that a high energy fuel source is used, the third
system-specifc technological area which must be ad-
dressed is PROPULSION, that is, how one converts the
fuels stored energy into useful, controllable work. This
involves some sort of engine, and a propulsor system.
The approach described in this paper is to use a chemical
fuel source driving a specialized scalable engine known
as the reciprocating chemical muscle (RCM), coupled
to fapping wing propulsors. This combination is deemed
to be optimal for indoor MAV missions where the MAV
is more than a simple fying machine, but a robot capable
of demonstrating various insect-like behaviors including
the ability to land, crawl, and take off again.
Form Follows Function
Beyond the fact that every living thing capable of sentient
navigation employs fapping wings for sustained aerial lo-
comotion, certain features of fapping wing fight make it
attractive for those missions in which micro air vehicles
are believed to have the greatest potential.
Many missions for micro air vehicles (MAV) have been
proffered, but all basically fall into the categories of out-
door, urban, and indoor. The domain for MAVs will
be as key elements of indoor missions. Major, and perhaps
insurmountable obstacles confront MAVs that fall prey to
the forces of the environment. Wind and rain can prevent
outdoor MAV fight from taking place as the tiny air vehicle
could expend its entire energy store getting nowhere in an
attempt to fy at 20 kph in a 20 kph head wind. Similarly,
rain will not only attenuate signals from the necessarily
high frequency command links but may even push the tiny
craft to the ground. Besides, assets exist for most outdoor
reconnaissance missions why use a MAV?
Proponents would argue that MAVs put the reconnaissance
potential in the hands of the users that need specifc infor-
mation in a timely manner. Perhaps a better solution would
be to invest in networked communications systems that
can get the same information to the foot soldier in a timely
manner from existing unmanned aerial vehicle (UAV) as-
sets such as Predator or the Global Hawk. Global Hawk
will look over all hills in the theater of war, providing con-
tinuous 0.09 square meter (1 square ft) resolution views of
the ground from an altitude of 20 km (65,000 ft) for periods
of up to 36 hours! Multiplexing the Global Hawk sensors
to take snap shots of specifc regions of the battlefeld and
to deliver them to individual users on the ground in near real
time is probably an easier and better integrated approach
to C
3
I than the anarchy of hundreds of tiny personal eyes
in the sky careening at the mercy of the wind.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
Urban settings, where the next generation of conficts are
predicted to occur, present diffculty for existing UAV as-
sets. This is because most UAVs are fxed wing vehicles
and are too fast to negotiate the urban canyons. Flying high
over a city is of use, but if one could gather reconnaissance
down in these urban canyons between buildings, then a
greater situational awareness could be had. MAVs are a
reasonable candidate for this mission since they are smaller
and potentially slower than conventional UAVs. Even fxed
wing MAVs could conceivably negotiate city streets, but
MAVs capable of slow fight and even hover would afford
the ability to stop, look into windows, or even land in tight
spaces to place sensors. On the other hand, wind and rain
will still plague these tiny air vehicles, and the occlusion
of signals by buildings will exacerbate communication
and navigation.
The real mission niche for MAVs will be indoors where
the environment is controlled, and there are no existing
airborne reconnaissance craft that can negotiate hallways,
crawl under doors, or navigate ventilation systems in an
attempt to complete a reconnaissance mission. It is the
indoor mission that will ultimately justify the development
expense. The very nature of an indoor mission will ne-
cessitate (1) multimode vehicles (fying/crawling/rolling),
and (2) autonomous navigation. These two features of
an indoor MAV are not absolutely necessary for outdoor
missions, but outdoor MAV missions are themselves not
absolutely necessary. Therefore, investment in the design
of autonomous multimode MAVs which incorporate these
features from the inception of their design is paramount.
Why Flapping Wing Flight?
If the most justifable missions for MAVs are indoors, then
a vehicle must be optimized to negotiate constricted spaces
that are bounded on all sides, land and take off with min-
imal ground roll, and circumvent obstacles (e.g., doors).
Fixed wing solutions are immediately discounted because
they require either high forward speed, large wings, or
a method for creating circulation over the wings in the
absence of fuselage translation.
High speed is not conducive to indoor operations because
it results in reduced reaction time, especially when au-
tonomously navigating through unbriefed corridors or amid
obstacles. When indoors, slower is better.
If, on the other hand, the wings are enlarged to decrease
wing loading to accommodate slower fight, the vehicle
soon loses its distinction as a micro air vehicle. Current
wisdom defnes a micro air vehicle as having no dimension
greater than 15 cm. Even at this scale, the forward speed
required for a fxed wing vehicle to effciently stay aloft
violates the criteria for negotiating constricted spaces..
Finally, there are methods for creating circulation over the
wings in the absence of fuselage translation. This can be
done by blowing the surfaces of the wing to increase lift
in an intelligent manner by using an internally-generated
pressure source. This has been demonstrated in manned
aircraft and certain experimental unmanned vehicles,
but is typically ineffcient unless there is a source of gas
pressure already available (such as bleed air from a gas
turbine engine).
Another way to move air over a wing without fuselage
translation is to move the wing relative to the fuselage and
the surrounding air. This can be a circular motion as in a
helicopter rotor, or it can be a reciprocating motion as in
a fapping wing. Both serve to create a relative wind over
an airfoil thereby creating lift.
A rotor is mechanically simple to spin, but does not use all
parts of the wing (rotor) with the same effciency since the
inner section near the rotor hub moves more slowly than
the tip. The same thing can be said for a fapping wing
where the greatest relative wind is created at the wing tip,
and none at the root.
A signifcant advantage of a fapping wing over a rotor is
the rigidity of the wider chord wing relative to the high
aspect ratio of a narrow rotor blade, and the fact that it
can be fxed relative to the fuselage (e.g., nonfapping
glide) to reclaim potential energy more effciently than an
autorotating rotor.
It could also be argued that a fapping wing implementation
is an inherently lower bandwidth system than one using a
helicopter rotor. Both systems require cyclic (once-per-fap
or once-per-revolution) control inputs to maintain vertical
lift and stability, but the frequencies at which these inputs
must be generated can be much lower for comparably sized
fapping implementations.
There is also a stealth advantage of a fapping imple-
mentation over a comparably sized rotor design in that
the acoustic signature will be less because the average
audible energy imparted to the surrounding air by the beat-
ing wing is much less than that of a rotor. The amplitude
of vortices shed from the tips of the beating wing grows,
and then diminishes to zero as the wing goes through its
cyclical beat, whereas the rotor tip vortices (which are the
primary high frequency sound generator) are constant and
of higher local energy. The sound spectrum of a fapping
wing will be distributed over a wider frequency band with
less energy occurring at any particular frequency, thereby
making it less noticeable to the human ear. All the energy
of the rotor spectrum will be concentrated in a narrow band
that is proportional to the constant rotor tip velocity.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
As the diameter of a rotor system decreases with the size
of the air vehicle design, it will become less effcient since
the velocity at the tips will decrease while the useless center
portion becomes a larger percentage of the entire rotor disk.
To compensate for this, the designer will tend to increase
the rotation frequency of the rotor to maintain lift for a
given fuselage mass and power source. The increased
rotation frequency will increase the frequency and energy
content of the sound produced.
On the other hand, as the wing span of a fapping wing
system is decreased, wing beat frequency must similarly be
increased to maintain lift for a given fuselage mass, but the
spectrum of the sound produced will simply broaden with
more energy occurring at higher frequencies. Though the
work produced to lift the fuselage mass may be the same
as that for the rotorcraft, the energy will be expended over
a wider acoustic bandwidth, but unlike the rotorcraft, it will
be nonuniformly distributed in the horizontal plane. The
net result is that a any fapping wing approach will be less
noticeable than a rotary wing approach because the sound
spectrum produced will approximate wide band white noise
rather than a discrete tone.
The fapping wing is conducive to slow fight and even hov-
er. It allows for short take off and landing, and may have
advantages over other techniques in terms of its acoustic
signature. All of these features are desirable for indoor
operations, but what about circumvention of obstacles such
as doors? None of the techniques mentioned so far has any
particular advantage when it comes to movement through
small openings such as partially-opened doors or under
closed doors. Similar problems exist for small openings
like windows, air vents, and pipes.
The solution is to have a multimode vehicle that is capable
of not only fight, but ground locomotion. Crawling is not
a particularly effcient form of locomotion if large distances
must be traversed, but a machine capable of only fight is
effectively neutralized were it to encounter a closed door.
If a fying machine could drop to the foor and crawl the
small distance necessary to go under the door, then the
mission could continue.
The notion of a hovering humming birdlike sensor plat-
form that darts about a room inspecting different items
of interest, is constricted in the near term by the energy
density of its power source. Until greater power densities
can be achieved, the likely mode of operation will entail a
covert quick entry to a distant area using fight, followed
by a precise positioning of a sensor using ground loco-
motion. This may represent one percent of the overall
mission. The remaining ninety nine percent will revolve
around the operation of the emplaced sensor from its re-
mote vantage point.
Flight Concept Under Development
Various wing beating concepts have been analyzed by the
authors and modeled kinematically using the computer
aided design (CAD) tool, IDEAS. The obvious fapping
mechanism of two opposing wings used by most natural
fiers was ultimately discarded in favor of an X-wing design
incorporating two sets of wings.
The advantage of the our X-wing concept is that it can be
conveniently scaled to micro sizes because it is mechani-
cally simpler than the birdlike fapping mechanism. The
use of four wings is necessary to resolve moments about
the fuselage, but also adds longitudinal (pitch) stability to
the vehicle. A built-in dihedral in each wing pair provides
a degree of lateral (roll) stability.
An X-wing glider has been built and has demonstrated
stable roll and pitch qualities at the milli scale where
the initial internal research is focused. Our milli scaled
testbeds are being constructed with wing spans on the order
of 30 to 46 cm (12 to 18 inches). This rationale comports
with the reasoning of Ellington who states that sizing
constraints for near term fying robotic insects sug-
gest that larger machines would be the best starting point.
Maximum speeds would be low, but because of their larger
size, fabrication would be easier and they would offer a
more convenient testbed for the development of control
systems. (Ellington
4
, 1997). All designs are being done
with scalability in mind. Many concepts that were work-
able at the milli scale have been abandoned because they
would not scale to the micro level as easily as others.
A kinematically-correct X-wing fapper was constructed
for evaluation (see Figure 1) and now resides on display
at the Museum of Victoria, Australia. This vehicle was
designed to move with opposing wing motions in which
the forward wing pivots in one direction while the aft wing
pivots in the opposite direction. The net result is a bal-
ancing of forces, with two wings always in an up-stroke,
and two wings always in a down-stroke. The wings and
mechanism in Figure 1 are not what will be used in a fying
version, but only demonstrate the X-wing kinematics.
Flapping Dynamics in the Insect Kingdom
Four degrees of freedom in each wing are used to achieve
fight in nature: fapping, lagging, feathering, and span-
ning. This requires a universal joint similar the shoulder
in a human. Flapping is an angular movement about an
axis in the direction of fight. Lagging is an angular move-
ment about a vertical axis which effectively moves the
wing forward and backward parallel to the vehicle body.
Feathering is an angular movement about an axis in the
center of the wing which tilts the wing to change its angle
of attack. Spanning is an expanding and contracting of
the wingspan.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
Not all fying animals implement all of these motions. Unlike
birds, most insects do not use the spanning technique. Insects
with low wing beat frequencies (17-25 Hz) generally have very
restricted lagging capabilities (Brodsky
3
, 1994). Insects such
as alderfies and mayfies, have fxed stroke planes with respect
to their bodies, and the only way these insects can alter the
stroke plan with respect to gravity is to change their body angle
(Brodsky
3
, 1994). Thus, fapping fight is possible with only
two degrees of freedom: fapping and feathering.
Using only these two degrees of freedom, there are 3 important
variables with respect to wing kinematics: wing beat frequency,
wing beat amplitude, and wing feathering as a function of wing
position. When coordinated, these motions can provide lift not
only on the down stroke, but also on the up stroke. The abil-
ity to generate lift on both strokes results from a change in the
angle of attack of the wing whose tip inscribes an ellipse when
considered relative to a body-referenced point. The ability to
generate lift on both the up- and down-stroke leads to the po-
tential for hovering fight in entomopters and ornithopters.
Wing beat amplitudes vary in nature from approxi-
mately 25 to 175. In general, as wing beat frequency
increases, wing beat amplitude decreases. The feath-
ering of the wing as a function of wing position is
crucial to the fight dynamics. Each line represents
the wing section at some arbitrary position across
the span of the wing (Ward-Smith
5
,1984). Generally
insects with a constant, vertical stroke plane must use
a large angle of attack on the descending part of their
wing trajectory.
Other techniques such as optimizing wing shape,
using elastic wing deformation, and employing the
Weis-Fogh clapping mechanism (Lighthill
6
, 1975)
can be used to enhance the wing kinematics, and thus
produce more effcient fapping fight.
Entomopter X-Wing Flapping
Like the alderfies and mayfies, the entomopter will
have a fxed stroke plane for each of its four wings.
Coupled to the RCM, each wing pair will be part of
resonant mechanical structure that will provide a self-
regulating wing beat frequency. The amplitude of
the wing beat is a function of the stiffness and spring
constant of this structure. The third important vari-
able, feathering, is accomplished through the use of
smart materials that exhibit a different compliance
under varying loads.
This latter feature will be controlled by not only by
the wing rib structure, but also the interstitial wing
material itself. Stereolithography and Fused Depo-
sition Modeling techniques have allowed the design
team to create intricate wing structures directly from
computer models. Careful attention is being paid to
material selection. Resilience, stiffness in opposite
planes, chemical compatibility, and ease of bonding
are but a few of the points being considered in the
choice of wing materials. Figures 2 through 4 show
wings being grown in our stereolithography machines
as well as ABS wing stiffening structures with, and
without interstitial materials.
These wing structures are designed with hollow micro
passages to allow intelligent venting of waste gas
from the RCM over the wing surface for directional
control of the entomopter. The circulation controlled
airfoils of the entomopter allow differential modu-
lation of the lift while maintaining a constant auto-
nomic wing beat. This simplifes the mechanics of
the wing and is scalable to the micro level by using
valves constructed with microelectromechanical
systems (MEMS) techniques.
Figure 1. X-wing entomopter kinematic model.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
Figure 2. Wings grown using
Stereolithography.
Figure 3. Wing rib structure pro-
duced
through Fused Deposi-
Figure 4. Fused Deposition Mod-
eled
wings with interstitial
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
Power to Fly
The power necessary to achieve fapping fight can be
calculated by using formulas derived by Azuma
2
, 1992.
This power is mainly a function of the following vari-
ables: vehicle mass, fapping frequency, forward speed,
wing chord, wing span, and wing beat amplitude. Example
calculations for a vehicle weighing 50g and having an ideal
100% effcient RCM have been estimated (Michelson
7
,
1997). Based on this analysis, just over a watt of power
would be necessary to propel such an entomopter. Weight
reduction is the most critical factor in creating a successful
entomopter. The equations of fight contain terms in which
weight contributes to the fourth power. Note that a dou-
bling the entomopter mass from 50g to 100g results in
almost eight times the required power. For this reason it is
critical that entomopter structures serve multiple purposes.
As an example, wings could also be antennas, legs could be
inertial stabilizers in fight perhaps someday the fuselage
might even be itself a consumable fuel source!
Reciprocating Chemical Muscle
The Reciprocating Chemical Muscle is a mechanism that
takes advantage of the superior energy density of chemical
reactions as opposed to that of electrical energy storage
which is the approach currently being taken by most other
MAV researchers. For example, the energy potential in one
drop of gasoline is enormous compared to that which can
be stored in a battery of the same volume and weight.
The RCM is a regenerative device that converts chemical
energy into motion through a direct noncombustive chem-
ical reaction. Hence, the concept of a muscle as opposed
to an engine. There is no combustion taking place nor is
there an ignition system required. The RCM is not only
capable of producing autonomic wing fapping as well as
small amounts of electricity for control of MEMS devices
and the nervous system of the entomopter, but it creates
enough gas to energize circulation-controlled airfoils. This
means that simple autonomic (involuntary, uncontrolled)
wing fapping of constant frequency and equal amplitude
can result in directional control of the entomopter by vary-
ing the coeffcient of lift (C
L
) on each of the wings, thereby
inducing a roll moment about the body of the entomopter
while in fight.
Of particular interest to the RCM design team was the
correspondence between the amount of force that can be
generated by a reciprocating muscle and the maximum
frequency with which this force can be applied. Also, as
the frequency is increased, how much linear motion can
be achieved in a reciprocating device.
Initial tests using a nonfying RCM testbed to which simple
wing spars were attached (see Figure 5), demonstrated that
the reciprocating chemical muscle concept is valid. This
test bed was mechanically ineffcient, but was still able to
produce wing beat frequencies of up to 10 Hz in the rather
massive testbed structure.
The amount of power necessary for fapping wing vehicle
fight has been the subject of many articles, and has been
estimated for insect-based fying machines by both El-
lington
4
and Michelson
7
. Calculations based on empirical
data taken from the RCM testbed indicate that the power
available for fight is greater than the power necessary for
fight (Michelson
7
, 1977). From this, the force required
to move a wing can be estimated, as can the required wing
beat frequency for a given wing design. If the force avail-
able is suffcient, then the amount of linear motion achieved
by the muscle will become less important as any motion
can be amplifed through something as simple as a lever
arm, but with the attendant loss of force deriving from
the ratio of the mechanical disadvantage presented by the
motion amplifying mechanism used.
The frequency of operation is much more critical. Some
polymer, rheological, or shaped memory alloy muscles are
able to create motion, but not at the high repetition rates
needed for fight. As shall be discussed below, resonance
is critical to the energy balance of the fight system. It is
conceivable that slower acting muscles could pump energy
into a resonant system which is operating at a harmonic of
the fundamental muscle actuation frequency, but this will
require more force from the muscle each time it imparts
energy into the more rapidly fapping system.
The RCM is intended to operate at the same fundamental
frequency as the resonant wing structure. Some authors
have questioned whether energy is stored in the muscle of
an insect, or in portions of its exoskeleton (Alexander
8
,
1995). The RCM stores small amounts of energy in its
structure, but this is releasable at a higher frequency than
fapping will occur in the milli-scaled entomopter. The
predominant energy storage medium in the milli-scaled
entomopter is clearly the fuselage/wing structure, and
the resonance of this structure is what will determine the
fundamental resonant frequency of the entire system. The
RCM is self regenerative but its speed of reciprocation is
governed directly by the resonance of the fapping struc-
tures and not the muscle itself. So long as the muscle
response is faster than the resonant wing beat, then energy
will be imparted to the system at the right time, including
that energy stored temporarily within the muscle itself.
Ellington has tabulated estimates of the wing beat fre-
quencies necessary to support various masses in a hovering
entomopter. He has made certain assumptions about the
physical properties of the entomopter such as a peak-to-
peak wing beat amplitude of 120 degrees (without Weis-
Fogh clapping), wing aspect ratios on the order of 7, and
a mean coeffcient of lift (C
L
) in hover of 2. From this
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
analysis (Ellington
4,
1977) one can extrapolate that a milli-
scaled entomopter weighing approximately 50g could fy
with a wing span of approximately 25 cm (10 inches) and
a wing beat frequency of 20 Hz. In reality, the wing beat
frequency might only be between 10 and 20 Hz when the
double X-wing design of the entomopter is considered.
One question remaining to be answered is how much
motion should a muscle typically produce in order to eff-
ciently move a fapping wing? Fortunately, the literature
reports at least one benchmark test in which the wing
muscles of wasps were observed through an opening cut
in the cuticle of the thorax during fight. This experiment
revealed that contraction and expansion of the muscle fbers
only accounted for two percent of the overall muscle length
during each full beat of the wing (Gilmour
9
, 1993).
Another benchmark measurement has been reported in
the literature in which the muscle effciency and level of
resonant energy storage has been estimated for the fruit
fy Drosophila hydei. In this study, the mechanical eff-
ciency of the fight muscle was determined to be only ten
percent, while the energy stored elastically for resonant
release was estimated to be somewhere between 35 and
85 percent (Dickson
10
, 1995).
Bench Test Performance Comparisons
for the Milli-Scale RCM
The next step in the development of the RCM was to
reduce it to a size that would be compatible with the
milli-scaled entomopter. The 7.62 cm (3 inch) milli-
scale muscle shown in Figure 6 was constructed as both
a test article and for incorporation into the milli-scaled
entomopter.
Both frequency sweeps at fixed operating pressures and
variable pressure tests were conducted. Figures 7 and
8 show some of the results obtained.
Figure 7 is a graph comparing the force available to do
work versus linear travel in inches for different oper-
ating pressures at a frequency of 10 Hz. The maximum
pressure that the RCM can accommodate is merely a
strength of materials question. For convenience, the
milli-scaled entomopter is being designed to operate
in the 40 psi pressure range. Increasing the pressure
has several effects. First, higher operating pressures
are less fuel efficient. Second, the pneumatic stiffness
of the muscle is greater and hysteresis is minimized.
Third, the range of travel increases for higher operat-
ing pressures.
Figure 5. Reciprocating Chemical Muscle Test Bed.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
Of primary interest is the relationship between fuel use
(endurance) and the benefts derived from operating a
higher pressures. Figure 7 shows that doubling the pressure
provides four times the force available to do work while
approximately doubling the linear travel of the muscle. For
this discussion, the minor hysteresis differences between
30 psi and 60 psi will be ignored.
As mentioned earlier, certain wasp muscles only move on the
order of two percent of the overall muscle length during each
full beat of the wing. The empirical data shown in Figure 7 in-
dicates that the 7.62 cm (3-inch) RCM is able to provide travel
ranging from three to six percent of its length for pressures
between 30 and 60 psi. This relatively large range of motion
is accompanied by signifcant forces of approximately 2 to 6
pounds available to both the up beat and the down beat.
With this excess in available force to do work and a rea-
sonable range of motion, the benefts of higher operating
pressures become less attractive since twice the fuel will be
Figure 6. Milli-scaled Reciprocating Chemical Muscle.
consumed when operating the RCM at 60 psi than when it
is operating at 30 psi. The RCM structures can also be less
massive when designed for lower maximum pressures. So
greater endurance and less weight direct the design toward
lower operating pressures.
An important design consideration is the frequency of opera-
tion. As pointed out above, high fapping frequencies will
result in more aerodynamic lift for a given wing size. Since the
RCM is a reciprocating mechanical device, it has inertial limits
to its maximum useful frequency of operation at a given size
and mass. The inertia of the RCM moving masses necessarily
limits the linear travel of the muscle due to the time required to
accelerate those masses at mid stroke and decelerate them at
the maximum stroke excursions. This, of course will be true
for any reciprocating muscle. As the frequency of reciprocat-
ing motion increases, the travel distance will decrease while
maintaining an average position at the center of its range of
motion. This center will also correspond to the point of
greatest linear velocity.
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
For a reciprocating muscle of given mass and
force potential, increasing frequency will result
in decreasing linear travel. Figure 8 shows the
operational envelope for the milli-scaled RCM
based on empirical data collected at an operating
pressure of 40 psi. Note that the force to do work
in fapping the wings remains essentially constant
over the entire cycle, but as the frequency increases,
the linear travel of the muscle decreases almost
exponentially.
For a reciprocation frequency of 20 Hz, the max-
imum peak-to-peak excursion drops to about 0.05
inches (0.13 cm) which is only about 1.7 percent
of the muscle length. This is just under the value
reported for certain wasps as reported by Gilmour
and Ellington (Gilmour
9
, 1993), but it should be
noted that the corresponding power available to
do work is between 2.5 and 3 lbs! Because of the
high power available, a mechanical transmission
(fulcrum) of poor mechanical advantage can be
used to convert this small range of motion into a
large one while maintaining suffcient wing force
in both up and down stroke. To the degree that the
milli-scaled entomopter can be made to operate at
lower beat frequencies with larger chord wings,
the frequency of operation can be reduced, but the
average power available to do work drops as the
linear travel of the muscle is increased.
F
o
r
c
e

A
v
a
i
l
a
b
l
e

t
o

d
o

W
o
r
k

(
p
o
u
n
d
s
)
Linear Travel (inches)
6
5
4
3
2
1
0
1
2
3
4
5
6
-0.10 -0.05 0 0.05 0.10
Comparison of Available
Force for Different Muscle
Operating Pressures
(frequency held at 10 Hz)
30 psi
40 psi
50 psi
60 psi
UP BEAT
DOWN
BEAT
4

H
z
6

H
z
8

H
z
1
0

H
z
1
2

H
z
1
4

H
z
1
6

H
z
1
8

H
z
2
0

H
z
F
o
r
c
e

A
v
a
i
l
a
b
l
e

t
o

d
o

W
o
r
k

(
p
o
u
n
d
s
)
Linear Travel (inches)
3.5
3.5
3.0
3.0
2.5
2.5
2.0
2.0
1.5
1.5
0.5
0.5
0
1.0
1.0
UP BEAT
DOWN
BEAT
PROJECTION
OPERATIONAL
ENVELOPE
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
Comparison of Available
Force for Different Muscle
Operating Frequencies
(pressure held at 40 psi)
Figure 7. Comparison of Available Force for
different muscle operating pressures.
Figure 8. Comparison
of Available Force for
different muscle operating
frequencies.
Conclusions
Micro air vehicles are best suited to in-
door missions because the environment
is benign and no other assets exist to ad-
dress this area of reconnaissance. Indoor
operations will have to be autonomous due
to micro air vehicle size constraints that
prevent it from carrying various non-
scaling items such as lower frequency
transmission systems. Also, command
and control information can not be sent
through most steel-reinforced concrete
buildings with the required bandwidth to
allow for teleoperation of the vehicle.
When operating autonomously indoors,
micro air vehicles will have to be more
than air vehicles, they will have to be
aerial robots capable of multimode lo-
comotion that will include not only flight
13th Bristol International RPV Conference,
Bristol England, 30 March - 1 April 1998
but crawling. When in flight, they will have to be able
to move slow enough to negotiate winding corridors,
stairwells, and narrow openings. Slow flight for un-
obtrusive reconnaissance missions is best done with
flapping-wing propulsors.
Near term propulsion for tiny multimode robotic
vehicles will be fueled from chemical or fossil fuel
sources. Electrical storage density is insufficient to
support missions of reasonable endurance at this time.
A reciprocating chemical muscle (RCM) has been de-
veloped and tested at a macro- and milli-scale for use
in a mechanical insect called an entomopter. The
Entomopter uses a novel X-wing pair design that is
resonantly driven by the RCM.
Empirical tests on the milli-scaled RCM show that
it develops sufficient force and motion to drive the
wings of an entomopter at frequencies necessary for
flight. The characteristics of the RCM comport with
those of insects, though currently at a larger milli
scale. In particular, a muscle extension/contraction
range of 1.7 percent of the overall muscle length has
been demonstrated at a reciprocating frequency of 20
Hz and a force available of between 2.5 and 3 lbs over
the entire range of motion.
The design of the entomopter and its RCM have been
tailored with size reduction in mind, such that MEMS
implementations will be possible to further reduce size
and final production cost.
References
1. Ellington, C., The Aerodynamics of Flapping Animal
Flight, American Zoology, vol. 24, 1984, pp. 95 - 105
2. Azuma, A., Springer - Verlag, The Biokinetics of Flying
and Swimming, Tokyo, 1992, pp. 77 - 154.
3. Brodsky, A., The Evolution of Insect Flight, Oxford;
New York: Oxford University Press, 1994, pp 35 - 39.
4. Ellington, C., The Aerodynamics of Insect-based
Flying Machines, invited presentation at IROS-97,
Intelligent Robots and Systems Conference, Grenoble,
France, 1997 (from unpublished manuscript).
5. Ward-Smith, A., Biophysical Aerodynamics and the
Natural Environment, John Wiley & Sons, New York,
1984, pg. 93.
6. Lighthill, J., Mathematical Biofuiddynamics, Society for
Industrial and Applied Mathematics, 1975, pp. 179 - 195.
7. Michelson, R., Helmick, D., Reece, S., Amarena, C.,
A Reciprocating Chemical Muscle (RCM) for Micro
Air Vehicle Entomopter Flight, 1997 Proceedings
of the Association for Unmanned Vehicle Systems,
International, June 1997, pp. 429 - 435
8. Alexander, R.M., Springs for Wings, Science, Vol-
ume 268, 7 April, 1995, pp. 50 - 51
9. Gilmour, K.M., Ellington, C.P., Journal of Experi-
mental Biology, No. 183, pg. 101, 1993
10. Dickson, M.H., Lighton, J.R.B., Muscle Effciency
and Elastic Storage in the Flight Motor of Drosophila,
Science, Volume 268, 7 April, 1995, pp. 87 - 90