You are on page 1of 5

Imperial College London

Department of Aeronautics
3rd Year

L3 Application
Final Report

March 13, 2015


Group 8
Aditya Ranjan
Justin Tsui
Bruno Scalzo
Filip Karasiewicz
Umar Iqbal
Gui Fu
Pascal Gehlert

Table 1: Robot Statistics


Dimensions
Score
Size [mm]
Mass [g]
Jumping Distance [mm]

Values
2.74
70
27.48
132

Group 8

L3

March 13, 2015

Robot Design

A jumping robot was designed such that its merit function, the ratio of jumping distance to size, was
maximised. For this exercise, a design process involving simplicity and numerous iterations was
chosen.

Figure 1: Final robot.

Propulsion Mechanism

Different methods of providing propulsion to the robot were examined, including compressive and
torsional springs, magnetic coils and compressed gas. A compressive spring was chosen due to its
reliability, the range of compressive springs available commercially and the simple mechanism. This
allowed for a simple design and numerous iterations of designing, building and testing within a limited
time frame and constrained resources.

Spring and Motor Selection

Due to the wide range of commercially available compressive springs and the higher cost and narrower range of motors available, the motor was chosen first and the spring was matched to it. A series
of assumptions and calculations were performed to determine the required compression of the spring,
the mass and size of the robot, the launch angle and the merit function. This allowed for the required
spring constant and torque of the motor to be determined; k = 928 Nm2 , T = 0.0613 Nm. The lightest
motors were found not to provide the required torque. Heavier motors were not considered due to
the mass limit of 30 g. Instead of designing a gearbox to transfer a high revolution rate into greater
torque, a motor with an integrated gearbox was purchased. Weighing 10 g and with a gear ratio of
1000 : 1, it generated 0.5 Nm torque, allowing for the use of a stiffer spring.

Structural Calculations

Flexural stresses acted on the front plate due to tension from the carbon rods and wire, and a compression force from the spring;
3F L
= 3.7 MPa << (f of acrylic plastic = 100 MPa).
(1)
2bd2
Torsional stresses on the motor case were determined by approximating the motor case as a hollow
cylinder subjected to torsion such that
f =

max (D4 d4 )
16
D

(2)

Imperial College London

Tmax =

Group 8

L3

March 13, 2015

where e is the eccentricity and c the maximum distance from the neutral axis. From motor specifications, Tmax = 0.5 Nm and the inner and outer diameter of the motor case are , d = 0.012 m and
D 0.0122 m, respectively. Calculations showed a large safety factor for the motor case.
Lateral deflection of the carbon fibre rods, due to motor torsion was also considered. Using classical
buckling column theory, the motor case - carbon fibre rod system was modelled as a beam pinned on
one end and subjected to an eccentric load.
[
(
)]
2 EI
1
ec

P
Pcrit =
;
max = P
+
sec
(3)
L2
A
I
2 Pcrit
max = 70 MPa. A 600 MPa UTS of the carbon fibre implied a large safety margin.

Circuit Selection

The lightest possible circuit was designed. A 20 mAh LiPo Cell, 0.92 g, with a magnetic bahoma
connection to a R 51 DSM2 Ultra Micro Actuator Receiver, 0.30 g, was chosen, saving weight and
allowing for easy updates. A generic toy RC controller was used. For an ideal DC motor the voltage
applied is directly proportional to the torque it generates. The battery and receiver only operated with
3.6 V devices, and thus the motor only received this voltage instead of the 6 V it was designed for.
Assuming ideal behaviour, the motor applied a torque of 0.30 Nm, whereas the full compression of
the chosen spring required approximately 0.28 Nm of torque, leaving little room for error.

Foot Design

The centre of gravity of every component in global coordinates was used to determine the centre of
gravity of the robot. Assuming the model would be inclined at 45 degrees relative to the ground, the
foot was designed such that the centre of gravity lay directly above the point of contact with the ground
in both the compressed and uncompressed states.

Jump Simulation

A 4th order Runge-Kutta method was used to simulate the jump. Wings featuring 20 cm span resulted
in a 12 % increase in jumping distance, at a cost of 60 % of the merit function and were therefore,
deemed unacceptable. The calculated drag coefficient using CFD was 1.9, and the the theoretical
jump distance was 2.78 m.

Manufacturing

Key highlights of the mechanical implementation of the design included the light and efficient Kevlar
wire to constrain the motor casing, as the wire is only subject to tension, as well as the 3 mm carbon
rods to hold the motor casing in place, which resisted the shear stresses. The cam was laser-cut from
acetyl with a lower compressive strength at 90 MPa, but a higher toughness than acrylic. Components
including the foot and the shaft were 3D printed to eliminate inaccuracies by manually joining. The
epoxy double bubble was used to join components.
1st prototype The wire-frame structure from the initial designs was replaced with a main shaft, consisting of two parts, a middle and top plate, a motor casing, and carbon fibre rods. The geometry
was simplified to avoid stress concentrations. The subsequent shifting of the centre of gravity and
subsequent in-air rotation of the robot was deemed acceptable. This prototype had major structural
problems. The main shaft was susceptible to bending, as it was long and slender. The middle plate
was not able prevent the shaft and spring from bending out of plane.
2nd prototype In this design, the cam was used to compress the spring without using the shaft as a
proxy to mitigate the out-of-plane bending. The carbon rods could be shortened to reduce the size of
Imperial College London

Group 8

L3

March 13, 2015

the robot. Kevlar wire provided a balance force on the motor shaft, preventing the carbon rods and
the front plate from bending. A foot was introduced to maintain the chosen launch angle. It showed
promising performance, evenly though it rotated in air as predicted. Stress concentrations that were
predicted during a preceding finite element analysis showed to be the cause of the failure of the foot
upon launch.
Final Robot Various parts, such as the front plate were slimmed down to save on weight. The foot
was improved for greater stability and the joint of the foot and the I-beam was reinforced to prevent
damage upon launch. The foot was coated with tango black to increase the coefficient of friction with
dry concrete from 0.3 to 0.6 - 0.85.

(a) 1st prototype.

(b) 2nd prototype.

(c) Final Robot.

Figure 2: Evolution of the design.

Testing

The robot jumped 0.523 m in interim tests (18.8 % of theoretical distance), but only 0.132 (4.7 % of
theoretical distance), in the final test. This large difference in jumping distance is because the motor
stopped fully compressing the spring shortly before final tests. This may have been due to wear and
tear, stalling of the motor during trials, or damage due to the heat from soldering across the various
prototypes.
As the spring was not compressed to its full potential, the jumping distance was reduced. Additionally,
the robot rotated about the lateral axis during the jump. This was due to the shifting of the centre of
gravity in the first and second prototypes. One possible solution would have been to use two smaller
motors in a symmetric arrangement to keep the centre of gravity close to the lateral axis of robot.
Alternatively, the single motor could have been positioned in line with the thrust vector, either above or
below the spring, and worm gears could have been used to change the axis of rotation and compress
the spring. However, this would have been highly inefficient. Further research on this method might
prove fruitful.

10

Conclusions

With a larger timeframe available for the project, a lighter motor and a custom gearbox could have
been used. A fully customised spring with carefully chosen stiffness and dimensions would also have
offered a greater scope for optimisation of the robot size and jumping distance. As shown in the
costing sheet in the Appendix, a large proportion, 31.55 % of the total cost stems from shipping, as
components had to be purchased from different suppliers. Also springs could only be purchased in
bulk, raising the costs. Furthermore, a broken transmitter had to be exchanged and the modified
battery charger could not supply the correct voltage, hence further batteries had to be bought.

Imperial College London

Group 8

L3

March 13, 2015

Appendices
A

Costing Sheet

A summary of the parts ordered for this project.


Table A.1: Summary and Cost of Parts Ordered
Part

Cost []

Shipping Cost + VAT []

Total []

10 x Springs
Motor
Receiver
Transmitter
Charger
2 x Battery

6.95
16.51
23.33
16.67
0.99
4

8.28
6.25
6.76
6.76
0
3.5

15.23
22.76
30.09
23.43
0.99
7.5

Totals

Without Shipping []
68.45

Shipping []
31.55

With Shipping []
100.00

Imperial College London

You might also like