0% found this document useful (0 votes)
124 views13 pages

Mathematical Modelling of Doraemon

Uploaded by

any431243
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
124 views13 pages

Mathematical Modelling of Doraemon

Uploaded by

any431243
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

Downloaded from www.clastify.

com by 안안안

Introduction
Doraemon is a manga series and a name of a walking cat-shaped robot that came from the
future to change youth protagonist’s life using future gadgets that act in magical ways. It is
a nationwide known character in Japan and has been airing in Japan for more than 40
years.
As a citizen of Japan, I, too became fond of this series since when I was small. Although
now I understand that the technology probably would not be advanced while I am
Figure 1
alive, as a child I genuinely believed that his gadgets are possible sometime soon. It
was around that time that I started developing interests in industrial design¾ which
start off as a mere sketch of imaginary items that I wanted to make. Doraemon was a
huge source of inspiration in those ideas, and eventually led me in studying design principals by
myself. This interest is also reflected in my MYP Personal Project ‘Typography’ and taking SL Art in
Diploma Programme.
Since I aim to pursue a career in industrial design, I would take this math investigation as an opportunity to use
my skills in Adobe Illustrator, — a graphics editing software commonly used in an industry — that I have
developed throughout the years. In scope of finding the aim, this would support further development of my
career as an industrial designer by demonstrating my spatial perception ability as well as designing a unique

om
solution to the proposed issue (in this case finding the volume).

l.c
Aim
ai
gm
I would be able to utilise my skills in industrial design to simply steps, and to develop modeling skills by first,
creating Doraemon’s model graphically, then applying skills in mathematics of transformation and calculus
3@

(volumes of revolution) to find out his volume from limited information I am provided of – which is a skill that
is required for a professional designer.
24

By finding out the volume, it would allow to apply this skill when full-scale modeling, which is again an
31

important skill in assessing the ergonomics of a product. Silicon casting is typically a popular option in molding.
y4

The found volume would specifically be applicable in knowing the volume of silicon to be used in this process.
an

Splitting up into sections


ify

Before I dive in to calculations, in order to make this investigation intuitive, I have used a 3D Doraemon figure I
st

had since I was small. The 3D model I own is vertically symmetrical, which is significant for this investigation as
a

when using volume of revolution method, it will revolve the shape 2" radians, thus requiring the shape to be
Cl

symmetrical around the axis of symmetry.

Since Doraemon has a complex shape, it must be split into different sections. Three main section that comprises
his body are: Head (Face), Body, and feet. Exterior elements such as nose, arms, pocket that stick out of the main
body parts will be calculated individually and added on to the original volume, as these elements cannot be
integrated to the original equation.

To start off, the picture of Doraemon was calibrated on Illustrator so that it would have a
height of 10cm. In this process, a picture of Doraemon was taken straight above, and the
picture was scaled. The scaling is performed in order to provide ease when magnified to
the actual scale the creator of Doraemon specifies.

Defining the height would also signify that the whole of body elements across the x-axis
would add up to 10cm. Since I would be using multiple equations to recreate his body,
Figure 2 – Doraemon this would support in specifying the domain as well.
figure calibrated to
10cm height

1
Downloaded from www.clastify.com by 안안안

Basic Equations
Based on the adjusted height, I have created an over view of Doraemon’s body using Illustrator’s Ellipse tool and
Pen tool to visualize the outline of his shape. Ellipse tool can create shapes based of circle, and
Pen tool can draw any sort of lines. In both cases, the software will provide the dimensions (width and height) of
created shapes.

Green line is used for the head.


Blue line is used for the bump on
the neck.
Pink line is outlining his stomach
and leg area.
Orange line is used for his legs.

Figure 3 – A graphical representation of Doraemon using Illustrator

From this process, I was able to identify two main types of equation that can be used in order to recreate his body

om
structure. Since green and blue line were created using Ellipse tool and adjusted accordingly, we can here apply
transformation to the general formula of circle, where r is the radius:

l.c
#$ + &$ = ($
ai
gm
On the other hand, the shape of
pink and orange line resembles the
3@

shape of a logistic function (the red


box area of Figure 4). Although the
24

shape of pink line is opposite to that


31

of the logistic function, this can be


y4

solved through transformations. For


Figure 4 ¾ an example of Logistic the orange line, no problem is seen.
an

function Figure 5 ¾ the close up of orange and


pink line
ify

It may seem that orange and pink line also resemble ln(#) function too.
st

However as soon as I graphed, I realized that it is not suitable for modeling,


a
Cl

because & value kept increasing as # increased, while as seen in Figure 5 the
lines are rather straight after a while.
Overall, logistic function be used for its nature of being able to set limits for
plateau. The general formula is below.
Figure 6 ¾ ln(x) function

)
&=
1 + + × - ./0

Where ) is the horizontal asymptote, + and 2 being the constant.

2
Downloaded from www.clastify.com by 안안안

Domain

The values on Figure 7 are numbers showing


the width of each element. This would be used to determine the domain of a function. Table 1 is a summary of
what domain each function will be using.

Table 1

Part Domain
Head 0 ≤ # < 5.303
Neck 5.303 ≤ # < 5.840

om
Body 5.840 ≤ # < 9.384

l.c
Feet 9.384 ≤ # < 10
ai
gm
Figure 7 — Tracing of a Doraemon with width of each element indicated
3@

alculations of main parts


24

Head
31

As stated above, head will use the circle formula. To begin with, following basic transformations must be
identified.
y4

(# − ℎ9 )$ + (& − :9 )$ = ( $
an

While both ℎ and : indicates the coordinate of the circle’s center, ℎ translates the x-coordinate, and : translates
ify

the y-coordinate. In this case, we can see that the green line directly extends from the origin (0,0) (Figure 3/) and
st

does not require a vertical translation.


a

:9 = 0
Cl

The dimensions of this green line are also provided by the software that was used to create this model. Since the
graphical representation was created by adjusting the circle, semi-circle, the maximum height (y-axis) of this line
would provide us the radius.

(9 = 2.968 (given)

The value of @A is also same as radius, as we want the circumference of the circle to lie on (0,0), not anywhere
negative.

Overall, following is the equation of the head after substituting the values back in.

(# − 2.968)$ + & $ = 2.968$


Rearranging:
& = B2.968$ − (# − 2.968)$
= B8.809024 − (# $ − 5.936# + 8.809024)
= B5.936# − # $

3
Downloaded from www.clastify.com by 안안안

As previously mentioned, the software provides dimensions for the shape created. From this data, I am also come
to aware that the ellipse is not a complete circle, and it is transformed slightly to match the outline. According to
the software, width of the circle is 5.764cm, and height is 2.968cm.

Dividing the width by 2 will give enough information to find out the scale factor.

5.764 (width)
= 2.882
2

I now can adjust the width of the equation to 2.882 cm by applying horizontal compression.

2.882
scale factor of # = = 0.9710242588
2.968

Applying the transformation:


# # $
& = U5.936 V W−V W {0 ≤ # < 5.303}
0.9710242588 0.9710242588

From the calculation above, function for the head Z(#) is found.

om
Z(#) = B6.17# − 1.08# $ {0 ≤ # < 5.303}

l.c
Neck ai
gm
The bump of his neck, in context is a result of him wearing a choker with a bell. Since it is a choker made of
strings, it is fair to assume that the cross section is a perfect circle. Therefore, it does not require enlargement or
3@

compression.
24

Letting [(#) be the function of his neck, we know that:


[(#) = B(($ )$ − (# − ℎ$ )$ + :$
31
y4

According to the information the software provides, the value for radius is following:
an

($ = 0.27
ify

In order to allow multiple equations to seamlessly connect with each other, &-value at the domain limit of Head
st

equation was found and applied as a transformation so that two equations will not overlap.
a

Z(5.303) ≈ 1.610
Cl

0
]-(^_`ab ^(acdba^_ec f& g h
1.610
:$ = 1.610
The centre of the circle was also moved for the same reason:
0.27 + 5.303 5.5703
ie(_jec^ab ^(acdba^_ec f& g h=g h
0 0

ℎ$ = 5.5703

Final equation of the neck:


[(#) = B0.27$ − (# − 5.5703)$ + 1.610 {5.303 ≤ # < 5.840}

Body
While logistic function exists, it is much complicated than the circle formula to transform and match the shape.
Thus, manually adjusting the formula is both time-consuming, and human-eye is never accurate to begin with.

4
Downloaded from www.clastify.com by 안안안

I sought for a solution where I can use the existing known data that could be reused mathematically in a form of
equation. Initially, free mathematics software GeoGebra was used to plot data points on a photograph, but I faced
a problem where I was not able to calibrate the plot scale as I desired to, which resulted in exporting a dataset

that was nowhere close to the scale seen in


Figure 7.

As I proceed, I found a free software ‘WebPlotDigitizer’, on the internet. This software was developed aiming to
‘reverse engineer images of data visualizations to extract the underlying numerical data’ (Rohatagi A., 2018),
which suits perfect for my situation where I have a visual representation without numerical data, but with
existing scale.

om
After exporting a higher resolution image from Illustrator and imported into WebPlotDigitizer, I was able to
successfully calibrate so that the scale of x and y would match up to that of the image. Following this, I have

l.c
plotted points on the image along the pink line to uncover the numerical value of the body outline. The plots
were then extracted for further analysis. ai
gm
3@
24
31
y4
an
ify
a st
Cl

Figure 8 – Process of plotting red dots on the Pink line. Check Appendix A for raw data.

Datasets extracted were analysed in GeoGebra. This software will output the equation for the binary logistic
regression from datasets using =FitLogistics(Dataset) command.
A regression is a statistical method of summarising the finding between independent variables and dependent
variables. Although this case, there no such thing as variables. Despite this, the core aim for the regression is to
minimize the residual error (the difference between actual value and predicted value) when compared with the
actual data sets, and GeoGebra will aid in minimizing this.

5
Downloaded from www.clastify.com by 안안안

1st attempt
In Figure 9, the black dots are the extracted data and
orange line is the best-fit GeoGebra gave. As it is apparent
from Figure 9, the best-fit command did not work very
well. This is perhaps because as seen in Figure 4, a logistic
function show an exponential growth rather than
exponential decay.
The equation exported was following:
1.69957
&=
1 + 0ℯ l.m990
Figure 9 – Initial attempt on Geogebra The accuracy of this model was measured using the R-
squared (R2) value.
R is a statistical measure of how well the trend is being represented in a model, from scale of 0 to 1, where 0
2

being the least accurate and 1 being a perfect model.


R2 for Figure 9 was 0.786. The high R2 value despite the discrepancy between real data perhaps arises from the
data plots that are plotted horizontally.
The R2 for the plots located in between 8.5 < # < 9.5 in particular was 0.396, further indicating that the model
generated here is inaccurate.

om
2nd attempt
I assumed that this error could be because the shape Figure 5 had was mirrored compared to normal shape of

l.c
logistic function (Figure 4). Therefore, I performed a horizontal reflection for all the datasets extracted, and thus
would match the shape seen in logistic function. ai
gm
Additionally, the whole data set was moved to quadrant 1, as few of the plots after reflected invaded quadrant 4.
Following translation was applied:
3@
24

9.322465
n(acdba^_ec f& V W
0
31
y4

-9.322… was the smallest value amongst the data plots, and translation had to be greater than this to make
everything in quadrant 1. This point now lies at # = 0.
an
ify

After reflection and translation, GeoGebra was able


to give out an equation of a logistic model that fits
st

the dataset.
a
Cl

R2 value for this refined model was 0.995, in which


is a fair value to conclude that it is accurate enough
to be used.

Figure 10
Following is the equation generated by GeoGebra
1.72496
_(#) =
1 + 0ℯ .$.moop90

As I applied several transformations to obtain this equation, few additional steps must be taken to make this
model usable. First, the process of horizontal reflection must be reversed.

1.72496
Horizontal Reflection: _(−#) =
1 + 0.54605ℯ .$.moop9(.0)

6
Downloaded from www.clastify.com by 안안안

3.544 + 5.84 9.384


Translation by V W=V W
0 0

1.72496
_x−(# − 9.384)y =
1 + 0.54605ℯ .$.moop9(.(0.z.{m|))

Therefore, the final function ℎ(#) for his body is _(9.384 − #):

1.72496
ℎ(#) = {5.840 ≤ # < 9.384}
1 + 0.54605ℯ $.moop90.$}.lz9|9{m|
Feet
Similar procedure as Body section was used as well for the feet. WebPlotDigitzer allowed me to access the
numerical data points of orange line, and the datasets extracted from this software was imported to Geogebra,
which gave the following plots and model:

The R2 value for this function was 0.965, in


which is a fair value to say its accurate.

No transformations are required for this

om
model, as I did not perform any of them, as
plot resembled the shape of a logsitic

l.c
function from the beginning.
ai
gm
3@

Figure 11
24
31

Final equation:
1.39413
y4

:(#) = {9.384 ≤ # ≤ 10}


1 + (28255299675732400588 × 10{l )- .9|.p|9|o0
an
ify

Main components on graph


a st

Now that I am done finding 4 main equations that comprise the body, I have graphed and achieved following:
Cl

Figure 12

Unlike the graphical representation from Figure 3, above is mathematical representation. Therefore, integration
and volume of revolution could be used in order to find the volume of this shape.
É
$
]eb~- eZ (-Äeb~^_ec = " Å xZ(#)y Ç#
Ñ

7
Downloaded from www.clastify.com by 안안안

3D model wascreated based on these


four equations, and despite its few
lacking features (face, pocket, arms),
overall it looks faithful to the actual
design; thus modelled well.

Figure 13 – 3D model from the revolution of four equations

Volume of Revolutions
Head
o.{p{
]Ö = " Å Z(#)$ dx
p
o.{p{ $
= "Å VB6.17080# − 1.08068# $ W d#

om
p
o.{p{

l.c
= " áÅ ( 6.17080# − 1.08068# $ ) d#à
p
o.{p{ ai
o.{p{
gm
= " áÅ 6.17080# d# − Å 1.08068# $ d# à
p p
3@

o.{p{ o.{p{
6.187080# $ 1.08068# {
= " âä ã −ä ã å
2 3
24

p p
= "( [3.08540# $ ]o.{p{
p − [0.36023# { ]o.{p{
p )
31

= "( 3.08540 × 5.303 − 0.36023 × 5.303{ )


$
y4

= "(33.04667)
≈ 104
an
ify

Neck
st

o.m|p
[(#)$ dx
a

]è = " Å
Cl

o.{p{
o.m|p $
= "Å VB0.27$ − (# − 5.5703)$ + 1.610W d#
o.{p{
o.m|p
= "Å VB0.27$ − (# − 5.5703)$ + 1.610W VB0.27$ − (# − 5.5703)$ + 1.610W
o.{p{
o.m|p $
= "Å VB0.27$ − (# − 5.5703)$ W + (2 × 1.61) × B0.27$ − (# − 5.5703)$ + 1.61$ d#
o.{p{
o.m|p
= "Å (0.0729 − (# − 5.5703)$ ) + 3.21B0.27$ − (# − 5.5703)$ + 2.5921$ d#
o.{p{

Calculation for the highlighted section:

= Å 3.21B0.27$ − (# − 5.5703)$ d#

let ~ = 0.27$ − (# − 5.5703)$

8
Downloaded from www.clastify.com by 안안안

= 3.21∫ √~ d#
9
= 3.21∫ ~$ d#
d~
= −2(# − 5.5703)
d#
1
d# = d~
−2(# − 5.5703)
9 1
= 3.21∫ ~$ × d~
−2(# − 5.5703)
o.m|p
2 { 1
= 3.21 í × ~ ×
$ ì
3 −2(# − 5.5703) o.{p{
o.m|p
2 { 1
= 3.21 í × (0.27$ − (# − 5.5703)$ )$ × ì
3 −2(# − 5.5703) o.{p{
Substituting back to the original:

o.m|p { o.m|p
(# − 5.5703){ 2 (0.27$ − (# − 5.5703)$ )$

om
= " îä0.0729# − ã + 3.21 × ï × ñ
3 o.{p{
3 −2(# − 5.5703)

l.c
o.{p{

ai
gm
+ [2.5921#]o.m|p
o.{p{ ó
3@
24

= "x(0.419196845709 − 0.392954831739) + (−0.000008173599248 − 0.000221185027877)


31

+ (15.137864 − 13.7459063y
y4

= "(1.786696125871635)
an

≈ 5.61
ify

Body
st

z.{m|
a

]ò = " Å ℎ(#)$ dx
Cl

o.m|p
z.{m| $
1.72496
= "Å g h
o.m|p 1 + 0.54605ℯ $.moop90.$}.lz9|9{m|
≈ 30.5
Feet
9p
]ô = " Å :(#)$ dx
z.{m|
9p $
1.39413
= "Å g {l .9|.p|9|o0 h
z.{m| 1 + (28255299675732400588 × 10 )-
≈ 3.63
Total Volume for main parts

]öÉõú = ]Ö + ]è + ]ò + ]ô
≈ 104 + 5.6 + 30.5 + 3.63
≈ 144(3s. f. )

9
Downloaded from www.clastify.com by 안안안

Calculations of Exteriors
Hand
Using the dimensions given from the tracings on Illustrator from Ellipse tool, the volume of the hand is below.

1.693 $
#$ + &$ = g h
2
(#) = B0.8465$ − # $

Figure 14 – The dimensions of a circle: 1.693cm x 1.693cm

Volume of Revolution:
9.}z{
]ö = " Å (#)$ dx
p
≈ 2.54
Nose

om
Using Ellipse tool again, the dimensions of the red circle is:

l.c
width: 0.7761 cm and height: 0.8114 cm.
ai p.m99|
gm
The blue line goes through the center of the red circle. By setting the radius to , we
$
can find out the scale factor to create a circle with same dimensions.
3@

({ = 0.4057
24

Figure 15 – Tracings of 0.7761


31

the nose scale factor of # = 2


0.8114
y4

V 2 W
an

= 0.956494947
ify
st

& = B(({ )$ − (# − ({ )$
a

#
Cl

$
c(#) = U0.4057$ − V − 0.4057W
0.956494947
Figure 16 ¾ graph of n(x)

Volume of Revolution:
ù
Since nose is a semi-circle shape, $ is applied instead of " in volume of revolution.

" p.ll}9 $
cû = Å xc(#)y
2 p
≈ 0.143
Total Volume for exteriors
]ü0 = ]ö + ]ú
≈ 2.541 + 0.143
≈ 2.68 (3. s. f. )

10
Downloaded from www.clastify.com by 안안안

Total volume (main + exteriors)


]†°†É¢ = ]öÉõú + ]ü0
≈ 144 + 2.68
≈ 146(3s. f. )

Finding out the Volume when height is 129.3cm


Since ]†°†É¢ is a volume when height is 10cm, we can use scale factor : to find the volume when height is
129.3cm

129.3
:=
10
: = 12.93

]eb~- £`ab- §a`^e( (]£§) = (•_c-a( £`ab- §a`^e(){


(BBC, n.d.)
]eb~- ¶ℎ-c 129.3cm = ]†°†É¢ × (:){
= 316000 cm{ (3s. f. )

om
Conclusion
This investigation has fulfilled the aim of finding the volume of Doraemon by utilising my strength in

l.c
computer and graphic design into fields of mathematics. As a result, I was able to derive the approximate volume
of Doraemon when the height is 129.3cm. ai
gm
As an objective of a future industrial designer, not only appearance but also the materials that comprises
3@

the product is significant. In many ways, the industrial designer is responsible for the day-to-day interactions of
customers and designed product, therefore weight would be one of the factors that can alter the design of a
24

product. From the finding that his volume is 316000 cm3, and the official data of weight being 129.3kg, the
density is 0.4g/cm3. Comparing with 2.7 g/cm3 of the 2nd most common used metal Aluminium.
31

Furthermore, it is noted officially that he has a high-tech combustion engine inside his body, thus the
y4

density of whatever metal they use will be much lighter than 0.4 g/cm3, assuming that it is only used for
an

encompassing. Overall, it can be concluded that even if artificial intelligence technology evolves, it is unrealistic
to expect a robot like him in close future unless new strong and light alloy is developed to sustain his body
ify

elements.
st

Given that 1L = 1000cm3, the amount of silicon required would be 31.6L. The volume of silicon comes
a

in useful when casting a silicon inside the mould.


Cl

Visual Accuracy
Visual accuracy is compared as followed:

Figure 17

As seen in Figure 13, GeoGebra is not capable of adding details onto the 3D model it has displayed, thus I find it
hard comparing the visual accuracy for it does not look like Doraemon except the shape. Therefore, in Figure 17
I coloured the model manually by at the same time maintaining the scaling. As a result, from the aesthetic point

11
Downloaded from www.clastify.com by 안안안

of view it is fair to say that the equations have depicted the proportions and form of Doraemon accurately when
viewed from the side. This would also mark that I have successfully managed to create a 3D model out of a real-
life figure or a mere sketch; which is significant for an industrial designer, because the career requires numbers of
production of sketches, and modeling them on a computer before it turns into a real-life object.
Mathematical Accuracy
Although visually the model looks accurate, it remains unclear whether if it is faithful to the initial graphical
model created (Figure 3). I would be able to find this through applying the method I discovered while I
investigated ‘Body’ and ‘Feet’ section. Using WebPlotDigitier, I extracted points from the graphical model and
then compared the extracted dataset with the plots the equation will give, using R2 value. The result is following:
Table 2

BODY PARTS R2 Average R2 for all parts


Head 0.978 0.972
Neck 0.951
Body 0.995
Feet 0.965
As much as it looked accurate, the R2 value also show that it is also faithful to the graphical model I initially
created. However, as seen in the actual 3D figure in Figure 17, although Doraemon does resemble circular shape,
it is never a transformed circle, but a circle that is a bit inflated. While this is a source of error, it can also be seen

om
as a strength as it fully used the current mathematics knowledge to everywhere applicable. Additionally, high R2

l.c
value also proves that the steps I took are reliable and able to yield consistent results; which comes in significant
in future purposes. ai
gm

Other evaluations
3@

Limitations Effect Improvement


24

Arm, bell, pocket was not Decrease in overall volume figured out; Use a 3D figure with arms and feet
included inside the thus, less accuracy. Small elements might separated (less simplified model) to
31

calculation as 3D figure I seem negligible in 10cm model, but aim for higher accuracy.
y4

own did not seem to portray when scaled to 129.3cm it can affect in
the elements accurately. E.g. greater scale.
an

the side of arms were stuck


ify

to the side of the head which


made it impossible to
st

calculate the overall


a
Cl

dimensions of his arm.


The modelling was based on I wasn’t able to produce a mathematical By applying the improvement
side-view of the 3D figure representation which considered the stated above, this will allow me to
width when seen from the front. As a model from the front, as the
character, there are far more many times boundaries between hand and
that the character will be seen from front head is clear.
rather than the side.
The camera angle was not Causes a warp in proportion, thus Using a stand or anything that can
completely horizontal / shot making the modeling (and the tracing) stabilize the camera and figure
from the centre of the 3D inaccurate. As camera moves off from position so that picture for
figure (The picture used in the center, the proportions will be modeling could be shot
Figure 3) altered perpendicularly.

Moreover, this was my first time learning about logistic function. This knowledge was later applied on biology
lab report when investigating the limits on photosynthesis rate, and this function supported me in justifying
hypothesis.

12
Downloaded from www.clastify.com by 안안안

References

BBC. (2018). BBC Bitesize - National 5 Maths - Using similarity - Revision 3. Retrieved December 6,

2018, from https://www.bbc.com/bitesize/guides/zxmfmsg/revision/3

MathsIsFun.com. (2017). Circle Equations. Retrieved December 6, 2018, from

https://www.mathsisfun.com/algebra/circle-equations.html

Mueller, W. (n.d.). Logistic Functions. Retrieved December 6, 2018, from

http://wmueller.com/precalculus/families/1_81.html

Rohatgi, A. (2010). WebPlotDigitizer - Extract data from plots, images, and maps. Retrieved from

https://automeris.io/WebPlotDigitizer/

(TV Asahi). (n.d.). Doraemon [Hero image of Doraemon]. Retrieved from https://www.tv-

om
l.c
asahi.co.jp/doraemon/cast/img/doraemon.jpg
ai
gm
3@
24
31
y4
an
ify
a st
Cl

13

You might also like