You are on page 1of 4

SCI-1130 SPRING, 2014.

Arduino Controlled Catapult


Madeleine Fort, Austin Greene, Halley Pollock-Muskin, and Aidan McLaughlin

AbstractFor this project we simulated the motion of a


catapult and used the computational model to determine what
force is required to pull the swing arm of the catapult in order
to have a projectile land a specified distance away. A user can
input the coordinates of a target and our model can compute the
tension required between the swing arm and tension arm, the
rotation of the base plate, and the position of the stop arm to
fire a projectile to those coordinates. The smart catapult then
uses motors to adjust the tension arm, base plate, and stop arm.
Our simulations align with our expected behavior of the system.
When comparing experimental data to what our model predicts,
we found that the landing distances followed similar parabolic
patterns but the experimental data had shorter distances. Our
goal is to build a catapult that incorporates a computational,
experimental, and physics model. The computational model is
based on the analytical model and we want the computational
model to reflect the physical system. Our work also serves as an
educational resource for others. It is an example of a project that
incorporates hands on experimentation, use of microprocessors,
and physics modeling in a compelling fashion that could recreated
in a classroom setting.
KeywordsArduino, Catapult, Modeling and Simulation.

I. I NTRODUCTION
A. Purpose
For our project we are modeling the behavior of a catapult
and constructing it according to the design shown in Figure
1. The system we are creating is able to take a user input
distance and projectile mass and determine what launch angle
is appropriate. We are building this catapult as an exercise
in combining mechanical and electrical components into a
working system and are using a physical catapult to validate
our analytical physics model and computer simulation. The
work were doing can be useful to any one considering the construction of an Arduino controlled catapult, which has potential
educational applications, such as an introduction to embedded
systems (Arduino) or experience in system integration.
Our goals are to gain a better understanding of physics by
going through the process of creating a computational model
based off of our analytical model of the system, and then
actually creating the catapult to see how well we can model
the physical system both analytically and computationally.
B. Rationale and Approach
We began by studying what examples of Arduino controlled
catapults we could find online and decided to model our
catapult on the predominant design we found[10] , displayed
in Figure 1. None of the documentation we have encountered
measures the accuracy of the final product, so we intend to
report on the performance of our catapult by comparing it
to our simulations of the system. Another improvement we

Fig. 1. Catapult Design: different perspectives of the catapult design. An


Arduino modifies the position of the motors that control the system, and a
fourth motor (not pictured above) in the rotating platform allows for another
dimension of motion.

aim to make is providing for 2 degrees of horizontal freedom


rather than one as the designs seen online only have one.

C. Assumptions
The model ignores all forces that take energy from the
system. There is no air drag on the projectile post-release in the
model. We ignore friction in the catapult system, and assume
that the spring is linear. We also assume that the spring does
not hit itself when contracting. In reality, the spring is long
enough that, depending on the rotation of stop arm, it can hit
itself when contracting. We assume that the projectile hits the
ground and remains there instead of bouncing. The stop arm
and tension arm are presumed to not deflect or move in any
way once locked in place.
II.

D ESIGN AND C ONSTRUCTION

A. Catapult Design
Our catapult design is based on a pre-existing Arduino
powered catapult[10] . The catapult is constructed out of laser
cut micro-density fiberboard (MDF). The stepper motor will
have an attachment to hold the swing arm down made out
of bent metal. All motors will be controlled by an Arduino.
We are using servos to adjust the angle of the stop arm and

SCI-1130 SPRING, 2014.

tension arm and to hold the swing arm down. We chose servos
because they provide angular feedback. The tension arm allows
for auto-reloading. It can reduce tension in the spring and allow
the swing arm to be easily reset. See Figure 3 for a list of parts
used. The spring and the piece that holds down the swing arm
are both missing from the rendering, and the mounting plate
is missing.

spring 1 and spring 4 would suit our needs the best. We put
N
.
them in series to make a spring constant of 68.265 m
Mass
1.148 Kg
1.148 Kg
1.148 Kg
1.148 Kg
1.72 Kg

Spring
1
2
3
4
5

Rest Length
.0912m
.072m
.071m
.091m
.16m

Extended Length
.1743m
.1038m
.1118m
.1760m
.20m

TABLE II.

III.

Force Constant
N
138 m
N
367.9 m
N
323.83 m
N
135.06 m
N
425 m

S IMULATION

A. Software Design

Fig. 2.

Part
1
2
3
4
5
6
7
8
9
10
11
n/a*
n/a*

Diagram of catapult parts, outlined in Table 1

Description
Base plate
Bottle cap
Swing arm
Mounting plate
Servo motor
Stopping arm
Bar
Support beam
Tension beam
Eye screw
Side supports
Spring
Z bar

Material
MDF
Plastic
MDF

MDF
Steel
MDF
MDF
MDF
Steel

TABLE I.

Quantity
1
1
2
1
4
2
1
4
1
2
2
1
1

Notes
Servo underneath rotates this
Holds projectile
2 glued together
Clamped to a table
Arduino-controlled
Holds bar up, attached to servo
Halts progress of swing arm
2 on each side of base plate
Attached to servo, spring tensioner
On tension beam and swing arm
Servo located between the supports
Connects swing arm to tension arm
Holds swing arm down

* NOT PICTURED

We are developing two distinct programs. The first is a


Python physics simulation that uses the popular numpy and
scipy Python mathematics packages to solve the ordinary
differential equations of motion of the catapult and its
projectile. See Appendix A for the the equations we used to
model the system. The Python code takes as input the desired
distance the user would like to the projectile to land at and
calculates the right motor positions for the catapult. The
Python code uses an ODE solver to calculate the position,
velocity and acceleration at every time step. Using this
information, it finds the proper angle of the stopping arm and
the correct tension amount for the projectile to go the desired
distance. The program then calculates and outputs the distance
to turn each motor. This information is passed to a second
program written in Arduino C. The Arduino C program
moves the motors of the catapult to proper configuration. To
see our code, refer to Appendix B.

B. Specifying Materials
The first task in designing the catapult was determining
what spring constant would be appropriate for the scale of our
project considering what servos we are using. We made several
assumptions in our model. We assumed that the springs we are
using are linear. The catapult incorporates two springs in series,
and we assumed that the two springs have a motion equivalent
to one longer spring but with half the spring constant. The
model was built on the assumption that the spring does not
deflect. According to Hookes Law we can derive that
k = F/x
We checked and validated the spring constants of several
springs that we were considering. First, we labeled each of
our springs 1-5 in order to keep track of which was which
during the test. We found a projectile that weighs 1.148 kg to
test the springs with. Then we measured the rest length and
extension length of each spring using this known mass. At one
point, we had to increase the mass because one spring did not
extend using the first mass. See Table 2. Once this constant
was defined, we could model the rest of our system around
it. After experimentation, we decided that a combination of

B. Analysis
The Python program calculates how much the tension beam
rotation changes the spring length. We validate the model by
testing the actual trajectory versus the predicted trajectory.
We can predict the projectiles path based on the angle of
release and the force applied to the swing arm. Figure 5
shows the angular velocity over time. We derived this velocity
from angular acceleration. We solved for angular acceleration
using the moment of inertia and torque (see Appendix A).
We can find tension using Hookes law. The angle of release
is the angle between the stop arm and ground. Comparison
of the calculated landing distances and experimental landing
distances will show whether the model is accurate.

IV.

R ESULTS

We found after testing both the model and catapult, that the
two show minor discrepancies. See Fig 3.

SCI-1130 SPRING, 2014.

our assumption that drag is not a force that must included.


Additionally, it seems that we must include other systems in
our model such as the friction of the throwing arm on its
axle and the spring as it compresses. With these changes, our
model would be even more accurate and would better reflect
experimental data.

Effect of Stop Arm Angle on Final Position

3.5

Predicted
Experimental

3.0

Total X Distance (m)

2.5

2.0

1.5

1.0

0.5

0.0

0.51.4

1.6

1.8

2.0

2.2

2.4

Stopping Arm Angle (Rad)

2.6

2.8

3.0

3.2

Fig. 3. The shape of the experimental and predicted distances is similar.


The differences are accurate to what we expected to see between our model
without drag and the real world with drag. This data was collected as the
tension arm was held constant at 90 degrees

The difference between our predicted results and experimental results can be attributed to a few major factors. Primarily,
our model doesnt account for many retarding forces. The
model doesnt factor in drag, acting on either the projectile or
the catapult during its movement, friction between the catapult
arm and its axle, or the force absorbed by the stop arm as the
swing arm collides with it. All of these factors act to decrease
the energy of the projectile. If these forces were implemented
in the model, it would decrease the magnitude of the distance
the projectile would travel making the model more realistic.
Human error in experimental data collection, and the small
data set, could also attribute for the asymmetric shape of the
the experimental data.
A. Statistical Analysis
We measured a series of data points from our catapult and
compared it to our computer generated model. Our projectile
was a nylock nut weighing .003kg. Below is a table detailing
these measurements with the angles of the stop arm and tension
arm as well as the predicted distance and measured distance.
Stop
90
180
170
160
150
140
130
120
110
100

Tension
90
90
90
90
90
90
90
90
90
90

Predicted
1.65m
-.35m
-.089m
0.64m
1.68m
2.65m
3.21m
3.25m
2.79m
2.24m

Measured
2.06m
0m
-.24m
-.07m
.32m
1.18m
2.38m
2.88m
2.66m
2.13m

Difference
5.9%
-28.9 %
-787%
33%
93 %
25 %
14%
19 %
26%
6%

V. C ONCLUSIONS
This project was meant to describe the accuracy of our
computational model when considering a system such as our
catapult. As we can see, there is a slight difference between
our modelled distance and our actual distance. Our predicted
and experimental results are well within an order of magnitude
of each other. It seems likely that this discrepancy is due to

VI. N EXT S TEPS


One next step that was considered but not pursued, might
have been to add a sensing system in catapult which would self
determine the location of the target. The most important next
step is to incorporate drag and other retardant forces into the
computer model. At the moment, we do not consider the drag
of the projectile because the velocity and physical size of the
projectile are very small. With the addition of a wind sensor we
could factor in wind speed and direction which would allow
us accuracy outside as well as indoors.
A PPENDIX A
D ERIVATION OF A NGULAR ACCELERATION OF S WING A RM
We know that angular acceleration is equal to the Torque
on the system over the moment of inertia. The first equation
describes the acceleration at the end of the swing arm with the
torque on the swing arm from the spring and the moment of
inertia of the swing arm.
T
I
This is the equation for the moment of inertia at the end of a
rod. H here is the total length of the swing arm (13in). W is
the width of the rod or (.5in). m is the mass of the projectile.
r is the radius of the projectile. M is the distance from the
projectile to the pivot point.
=

h2 + w2
+ mr2 + mR2
3
Substitute force for torque. D here is the distance from the
from the projectile to the location of the applied force.
I=

h2 +w2
3

F DSin
+ mr2 + mR2

Hookes law. We are assuming our springs are idealized


springs. In retrospect, this in an area that could have been
improved upon.
F = kx
Plug in all the variables and simplify. This equation aids us in
finding the velocity of the projectile at the moment of release
so that we can determine its kinetic energy. We plug the
equation into the ODE solver and solve for and .
=

h2

3kxdsin
+ w2 + mr2 + mR2

A PPENDIX B
C ODE
A link to our code on github:
https://github.com/jagreene/SmartCatapult

SCI-1130 SPRING, 2014.

A PPENDIX C
ACKNOWLEDGMENT
Zhenya, Thank you for being an awesome professor. We had
laughs and physics and honestly, what more is needed. Brian,
you are a pretty chill Ninja.
We would also like to thank the machine shop workers who
laser cut our pieces for the catapult. The turn around time was
magnificent and for that, we thank you.
R EFERENCES
[1]
[2]

[3]

[4]
[5]
[6]

[7]

[8]
[9]
[10]

Arduino Catapult. CS252 Course Wiki Spring 2012 Sections 1 and 2


RSS. N.p., n.d. Web. 22 Feb. 2014.
Barista, Sinan. Investigating the Effect of the Swinging Arm Length on
the Horizontal Distance Travelled by an Object That Is Thrown from a
Catapult.Tedprints. N.p., n.d. Web. 22 Feb. 2014.
Brown, Douglas. Tracker Video Analysis and Modeling Tool for
Physics Education. Tracker. N.p., 2014. Web. 26 Apr. 2014.
https://www.cabrillo.edu dbrown/tracker/.
Catapult Physics Real World Physics Problems. N.p., n.d. Web. 21 Feb.
2014
cej12. Arduino Catapult. Instructables.com N.P.,n.d. Web. 22 Feb. 2014
Flueckiger, Patrick, and Roberts, David. Home (Physics of A Catapult.
Home (Physics of a Catapult).UNC at Chapel Hill, n.d. Web. 22 Feb.
2014.
Fong, Laurence, and Brian Self. Modeling the Dynamics of a Small
Catapult to Enhance Undergraduate Studies. ModelsandModeling.net.
California Polytechnic State University, San Luis Obispo, n.d. Web.
Ink, Winkle. Winkleink - Box of Wires. : Arduino Servo Catapult. N.p.,
n.d. Web. 22 Feb. 2014.
Serway, Raymond A. Physics for Scientists and Engineers with Modern
Physics. London: Fort Worth, 2000. Print
UnusualTravis. Arduino Controlled Catapult. Instructables.com
N.P.,n.d. Web. 22 Feb. 2014

You might also like