L3 Applications Exercise

Department of Aeronautics

Group 10
Alex-Hugh Wilson
Llewellyn Morse
Milan Patel
Samuel Jackson
Gabriel Wee
Mutian Li
Joseph Dudley

Academic Supervisor: Dr. Mirko Kovac

ABSTRACT
This report investigates our attempt at successfully replicating the jumping locomotion of an insect,
inspired biologically, in the form of a miniature robot. The robot was modelled to jump similar in
motion to a grasshopper. In order to create a prototype our initial design underwent many
modifications backed by numerous mathematical calculations and the analysis of 3D CAD models. Due
to a number of design problems, our final robot was unable to jump on the scheduled testing day. This
report evaluates what went wrong with our design, and suggests several future improvements that
could be made.

INTRODUCTION
In the natural world, animals have developed a variety of methods to help them navigate over complex
terrains. The grasshopper is a prime example, able to overcome obstacles longer than 20 times its size.
This feat is performed by the ability to store a large amount of potential energy in its elastic legs,
allowing it to propel forward with a large impulse. To mimic this ability to store elastic energy, two
torsional springs were compressed by a cam connected to a gearbox, which would propel the robot in a
way similar to the grasshopper. Once the mode of propulsion was selected, the robot was refined using
an iterative process to ensure the mechanical parts worked efficiently and effectively. The exact
method in which the robot achieves this is explained in detail below.

PERFORMANCE MODELLING
Our robot was designed to meet a series of design requirements, it had to be less than 20cm in size, less
than 30g in mass, actuated by an electric motor, remotely controlled, and could not lose mass during its
jump.
At the beginning of our design process we aimed for a set of design requirements that were more
challenging than the original set. We aimed to obtain a jumping distance of 1.5m, a maximum mass of
10g and a maximum size of 5cm. By using the following equations we obtained a required take-off
speed of 5.42m/s and a flight time of 0.39s for a jump distance of 1.5m.
vtcos(45) = 1.5

vsin(45) = gt

Using this take-off speed and a mass of 10g we obtained the take-off energy of our robot:
K E = 12101035.52 = 0.151J
In order to calculate the torque required by our springs we first needed to calculate the spring constant,
k, using the following equation:
U = K E = 12k2
Where U is the Torsion spring energy in J, k is the spring constant in Nm/rad and the angle the spring
turns in rad. Knowing that our launch angle was 45 we chose a value of 40 for so as to provide a
suitable safety margin. We obtained a spring constant of 0.620 Nm/rad or 3.89 Nm/turn. We then
calculated the torque of the springs:
T = k = 0.433N

DESIGN PROCESS
For our initial design we chose a jumping system similar to that of a grasshopper because it has a
relatively simple jumping system, which minimises the number of moving parts required in our design.
Our first design can be seen in Figure 1 and involved a system of gears powered by a small 0.4g,
0.88N/mm torque motor which turned a cam causing two legs to tilt backwards and two stainless steel
2098.8N/mm torque torsional springs to compress. The gear system was to be housed between two
frames made out of acrylic. The feet, the front leg and the cross-beams connecting the two frames
were to be made out of carbon-fibre rods. Our first design had a lot of problems, most noticeably with
its strength and toughness. The carbon fibre rods and the acrylic frames that traversed the mounting
panels, were too thin to support the large bending moments generated by the compression of the
torsional springs, and the main leg could have been made out of carbon fibre in order to reduce its
weight and increase its strength. As a result of this we created a 2nd design, which can be seen in Figure
2. This design featured a much greater use of carbon fibre as well as a thicker frame. The Aluminium
parts and the frame have both replaced with 3D printed versions made out of Endur plastic, which has
half the density of Aluminium and only 35% less strength than Acrylic. The frame is made out of four
separate parts that can be slotted together. Our 3rd and final design can be seen in Figure 3, it is very
similar to our 2nd design except that it is significantly lighter due to the removal of excess material.

MECHANICAL IMPLEMENTATION
In order to simplify the manufacturing and building of our robot we decided to 3D print the frame and
the joints, we also used 2 mm diameter carbon fibre rods for the legs, feet and for holding the 3D
printed parts together. In total, the design included 18 3D printed parts, 11 carbon fibre rods of varying
length, a cam which was laser-cut out of Acrylic, 4 gears that were bought online and two 11.34mm
diameter springs, as well as an IR-sensor, circuit and motor. The gear-cam system involved 4 gears; 1
worm gear, 1 36-tooth gear and 2 60-tooth gears, this provided a theoretical gear ratio of 552:1 . The
motors torque was transferred to the gear system by the use of a worm gear.
In addition to the motor
and battery, the circuit contained a phototransistor, a pair of resistors, a transistor, and a capacitor (see
figure 4). The additional transistor served to amplify the current to the motor, while the capacitor
helped smooth the current flow.
The final weight of our design was measured to be 27.4g.

ROBOT PERFORMANCE
Unfortunately, the robot we had devised was unable to jump on the scheduled testing day. One of our
first problems was that our initial batteries did not provide sufficient current to the motor, we solved
this by replacing them with batteries with a higher discharge current. We then had a problem with a the
interference fit between the final gear and its carbon-fibre axle, which was not tight enough to prevent
the gear from slipping. To fix this, we glued the final gear and the cam to their axle in order to ensure
that torque was transferred between them without any slippage.
The built-in plastic shaft of the motor was very short, giving only a small surface area to bond to the
metal axle of the worm gear. The long moment arm from this connection point meant it was stressed
repeatedly, and eventually it failed in shear due to fatigue. This made the motor unusable as there was
no way to transfer torque to the gear train. An attempt was made to replace the damaged part of the
motor, but this resulted in the motor gears falling apart, so the motor became permanently unusable
and preventing the robot from operating.
There are many improvements we could make to our design, the main problems of our design and our
proposed improvements are listed in the following table:
Part
Springs
Robot
Casing
Torque
Transfer
Motor
Springs
Circuit

Design problem
Improvement
The motor was unable to provide enough torque to
Use springs with a lower spring
compress both spring and it was too heavy (9g)
constant
A significant amount of time was spent redesigning and
Use ball bearings or bushing to
reprinting the robot casing due to problems with
reduce friction between the
interference and friction with the gears
gear axles
The interference fit of the carbon fibre shaft loosened
Use a shaft with a variable
and adhesives were not effective
geometry to attach the cam
Plastic shaft of the motor was too weak for its
Use a motor with a stronger
application
shaft
The motor was unable to provide enough torque to
Use springs with a lower spring
compress both spring and it was too heavy (9g)
constant
Size of the circuit made it cumbersome
Use an integrated circuit
Table 1: Design problems and improvements

CONCLUSION
To conclude, although our final robot failed to meet its design requirements, we do not consider the
design process to have been a failure. The individual elements of the design worked well when tested,
but integrating them proved more difficult than expected, and we should have set aside more time for
testing and fixing compatibility problems. We are confident the design would have worked if we had
fixed these problems in time.

APPENDIX

Figure 1: The first version of our design

Figure 2: The Second version of our design

Figure 3: The third and final version of our design

Figure 4: The circuit containing the motor, battery, and remote activation mechanism