You are on page 1of 3

L3 Applications Group Exercise

Group 4
Ciuca, H. Chattha, S. Christie, J. Delpech, C. Chu, C. Cheung, J. Cong
Department of Aeronautics, Imperial College London
Abstract This report presents the design,
manufacturing and testing of a small launchgliding robot. The design employed a mix between a rear wheel drive vehicle (for ramp
launching) and a Glider (for aerodynamic performance). The prototype was powered by a
remote-controlled electric engine, had a total
mass of 46.9g and was capable of travelling a total
distance of 3m. The cost of manufacturing the
robot was 17.8. A mathematical model designed
to predict the performance of the prototype is
detailed and shown to match with experimental
results.

I. INTRODUCTION
The field of Aerial Robotics has been under
continued development since the advent of military drones. Although this type of robot has been
most widely employed in warfare, the potential
for civil applications is endless. One example
would be rescue-type (search and find) missions
and our robot is designed to reach inaccessible
and remote areas using both terrain and launchgliding locomotion. The design requirements can
be found in reference ([1]). The key performance
parameter in our design is the distance between
the base of the ramp and the landing point.
II. PRELIMINARY DESIGN
A. Wing Design & Aerodynamics
A conventional glider design was chosen due
to favourable stability characteristics, high atL
tainable D
and plethora of research information
available ([2],[3],[4]). Nonetheless, wing design
was a challenge, due to a very low Re (Re
16, 000 assuming v = 3ms1 and c = 8cm), and
the fact that the produced lift needs to balance
the weight at the onset of the gliding phase. High
L
D was first established as a key characteristic.
However, the lift generated (for the same assumed velocity) was ten times lower than the
weight. Hence, the wing area was increased
at the expense of an increase in drag. From
the experimental study of a 2D cambered flat
plate at Re 20, 000 ([5]) and using the aerodynamic code QBlade, it was established that a
cambered flat plate allows an increase of 28%
in CLmax relative to the low Re NACA 8805.
Recognising Nature as the pioneer of aerodynamics, we based our wing design on a bat
wing profile (see Figure 2). This was due to the

amazing low-speed aerodynamic performance


of such creatures ([6]).
B. Vehicle Dynamics & Structural Aspects
Keeping the drive train simplistic and lowweight, an in-line gear system, sized for a gear
ratio of 2 : 1, is used to drive the rear shaft. This
has the added benefit of minimising transmission losses. The design aim was to minimise
mass whilst ensuring that the robot maintained
rigidity and stiffness during each test, although,
an overly rigid design would be too brittle to
survive a crash impact. For the initial chassis
design, a basic platform with cut-outs for the
wheels, was manufactured using 3mm Acrylic.
Testingof the acrylic components displayed possible areas of stress concentration. Reducing
the thickness of Acrylic to 2mm and rounding
all sharp corners immediately reduced the mass
of the part, and increased the impact tolerance
of the robot.
C. Stability
Longitudinal and lateral stability are key factors for the success of the mission. In this preliminary design stage only static stability was analysed by enforcing zero net moment around the
c.g. The only potential dynamic instability was a
short pitching oscillation, which would be heavily
damped ([2]). Lateral flight stability was ensured
through careful positioning of the c.g. on the axis
of symmetry. Longitudinal Static Stability plays
a vital role, especially in the transition phase
before gliding, where a good restoring moment
dCM
d < 0 is needed to avoid a nose-down dive.
This was ensured by having a large static margin
given by:
xc.g. lh Sh CLh
+
(1)
Kn =
c
cSw CLw
Where all symbols are as standard ([7]).
III. MATHEMATICAL MODELLING
A. Ground Phase
The A.C. of the body was assumed to coincide with the C.G. of the model, and the drag
and lift of the body were modelled empirically
([8]). The transmission loss was also modelled
empirically as a function of the number of gears
and complexity of the drive train and found to
be roughly 15%. The torque transmitted by the
engine was simulated in Matlab according to

the way the motor control was coded (either


a gradual throttle to full power in time t or a
step-input). The friction coefficient has been
determined experimentally. A Free Body Diagram of our model is shown in Figure 3. Lift
and drag coefficients for the wing and tailplane
were obtained from experimental results for flat
plates with corrections to account for 3D effects
and induced drag was modelled using LiftingLine Theory. A system of 5 equations: rotational
equilibrium of the rear and front wheels, horizontal and vertical equilibrium and sum of moments
about C.G., with 5 unknowns: f , F , Nr , Nf and
x was solved in Matlab to give the ramp exit
velocity. The program gave warnings if the front
or rear wheels would lift off (Nf < 0 or Nr < 0),
or if there would be slipping (F > Nr ).
B. Flight Phase & Optimisation
The longitudinal flight equations of motion
were modelled assuming that they can be decoupled from the lateral motion, and also assuming that the angle of attack varies passively
2
throughout the flight ( ddt2 0) due to the lack
of pitch control and powering during the flight
1
Sw
phase ([2]). Denoting = 2 m we arrive at ([2]):


g cos()

CL v
v

g sin() C v 2 v

D =

v cos()

y
v sin()

(2)

where the lift and drag coefficients are evaluated


for the whole model with reference to the wing
area. Obtaining the final distance travelled allows us to optimise our design accordingly. This
was done throughout the whole design process
to help us find the optimum engine choice (see
Table I), gear ratio and ramp angle. As our
design changed to account for manufacturing
considerations and test results, our mathematical model helped to refine our configuration as
seen in Table I and Figure 1.
IV. MANUFACTURING
The majority of components were laser cut
from acrylic sheets. Due to the 2D limitation of
the acrylic parts, interlocking joints were used
where possible. The laser-cutter had a tolerance
of 0.2mm, and hence interference fits were adjusted appropriately. The order of assembly was
chosen to minimise the number of operations required, ensure correct interlocking/spacing, and
reduce the amount of stress on the model during
production. Many parts were assembled using

epoxy glue. This added unneccesary mass and


decreased the impact resistance. Thus, many
parts were redesigned to house locking devices
for secure placement. Components such as the
motor, bearings, gears and wheels were bought
from Technobots as these parts relied on high
tolerances. The gears and wheels had twice
the mass quoted by the manufacturer, hence
many alterations had to be made to meet weight
requirements, including reducing the number of
wheels to 3, and reducing the chassis width.
V. RESULTS & DISCUSSION
The final designs flight trajectory is shown in
Figure 4. Unfortunately, on ground impact, the
model broke due to its frailty. The final testing
video can be found on our website [9]. Our
model showed potential by following a glide trajectory rather than simple projectile motion. The
robot locked onto a gliding path upon achieving
sufficient speed, but this happened too close to
the ground. The total distance travelled was 3m.
The experimental results are compared to our
mathematical prediction in Figure 5, and seen to
match well. Differences are due to the following
reasons: our glider did not jump straight, hence
our trajectory has an offset to the left in Figure
5, and the ramp exit velocity was lower than
expected. We believe that this is due to the
way voltage was supplied to the motor, rather
than due to the drivetrain, given its simplicity.
Nonetheless our mathematical model predicted
the total distance travelled extremely well, due
to the fact that we recorded our preliminary tests
and adjusted empirical factors and aerodynamic
coefficients accordingly.
VI. CONCLUSION
A promising design of a miniature launchgliding robot has been presented in this report.
The main features of our design and notable results are summarised in Table I. Improvements
should be aimed at preventing a slip angle in
the ramp motion, by careful positioning of the
front wheel, and improving the sturdiness of our
design. Furthermore, the idea of an even lower
aspect ratio wing, with a larger surface area
should be investigated. Given the mass limit,
this was not possible without a redesign of our
chassis. Other notable ideas present themselves
in a more careful study of how optimum voltage
should be prescribed to the engine, such as to
make use of its full power. The whole design process could have been streamlined by adopting
a design for manufacturing philosophy.

Figures and Tables

Comparison between test results and mathematical prediction

y (m)

0.8
0.6
0.4
Experimental Results (wing trailing edge position)
Ramp
Predicted trajectory with 3.1 m/s
Predicted trajectory with 2.8 m/s

0.2
0

0.5

1.5

2.5

x (m)

Fig. 1.

Fig. 2.

Final glider design: view from the side.

Final glider design: view from the top.

Fig. 5. Comparison between the mathematical model and


the actual test results of the manufactured robot. The figure
shows the flight path with the predicted ramp exit velocity
of 3.1m/s, as well as the one using the recorded velocity
of 2.81m/s.
TABLE I
The main characteristics of our design are summarised
here.
Final Distance travelled
Ramp-exit velocity
Engine Choice
Tstall
no load
Gear Ratio
Vehicle configuration
Launch ramp angle
Wing surface area
Wing span
Wing incidence
Wing m.a.c.
Tailplane surface area
Tailplane span
Tailplane m.a.c.
Tailplane incidence

3m
2.81m/s
Technobots Micro-Gear Motor
2oz in
6000rpm
2:1
3 Wheels-Rear-wheel drive
2.5
440cm2
40cm
6
10cm
51cm2
12.8cm
4cm
0

References

Fig. 3. Free body diagram of the mathematical model


considered in the ramp equations, with rear-wheel drive
shown here.

Fig. 4. Path trajectory of the gliding robot during testing


in the Flight Arena in the Imperial Aerial Robotics Lab.

[1] Mirko Kovac. L3 Applications 2015/2016. Department


of Aeronautics, Imperial College London, 2015.
[2] A. Vidyasagar, J.C. Zufferey, D. Floreano, M. Kovac.
Performance analysis of jump-gliding locomotion for
miniature robotics. Bioinspiratio & Biomimetics, Vol.
10, No. 2. March 2015.
[3] A.L. Desbiens, M.T. Pope, D.L. Christensen, E.W.
Hawkes, M.R. Cutkosky. Design principles for efficient, repeated jumpgliding. Bioinspiratio & Biomimetics, Vol. 9, No. 2. March 2014.
[4] A.L. Desbiens, M.T. Pope, F. Berg, Z.E. Teoh, J.
Lee, M.R. Cutkosky. Efficient Jumpgliding: Theory
and Design Considerations. Presented at IEEE, ICRA
2013.
[5] John McArthur. Aerodynamics of wings at low
Reynolds numbers. Presented to the Faculty of the
Graduate School of the University of Southern California, July 2007.
[6] A. Hendestrom, L.C. Johansson. Bat Flight: aerodynamics, kinematics and flight morphology. The Journal of Experimental Biology, No. 218:653:663, 2015.
[7] A.W. Babister. Aircraft Stability and Control. Second
Edition, Pergamon Press, 1961.
[8] S.F. Hoerner. Fluid-Dynamic Drag, theoretical, statistical and experimental information. First Edition, 1965.
[9] Group 4. Batglider [online] Available from:
http://batglider.weebly.com/

3.5

You might also like