TELKOM
NIKA
, Vol.14, No
.2, June 20
16
, pp. 655~6
6
4
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i1.2989
655
Re
cei
v
ed
No
vem
ber 8, 20
15; Re
vised
Febr
uary 8, 2
016; Accepte
d
February 2
7
, 2016
Quadrotor Path Planning Based On Modified Fuzzy Cell
Decomposition Algorithm
Is
w
a
nto*
1,2
, O
y
as Wah
y
u
nggoro
1
, Adha Imam Cah
y
adi
1
1
Departem
ent of Electrical En
gin
eeri
ng a
nd Info
rmatio
n
T
e
chno
log
y
, U
n
ive
r
sitas Gadja
h
Mada,
Yog
y
ak
arta, Indon
esia
2
Departme
n
t of Electrical En
gi
neer
ing,
Un
iver
sitas Muhamm
adi
ya
h Yo
g
y
ak
arta,
Yog
y
ak
arta, Indon
esia
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: is
w
a
nto.s3te
13@ma
il.u
g
m.ac.id
A
b
st
r
a
ct
T
he pur
pos
e o
f
this pa
per is t
o
pres
ent a
n
a
l
gor
ith
m
to
dete
r
mi
ne th
e sh
ortest path for
qu
adroto
r
to be abl
e to navig
ate in an u
n
know
n are
a
. T
he prob
le
m
in
robot navi
gati
on is that
a rob
o
t has inca
pab
i
lity
of findin
g
the s
hortest path w
h
ile
movin
g
to the go
al
p
o
siti
o
n
and
avoi
di
ng
obstacl
es. Hen
c
e, a mo
dific
a
ti
on
of severa
l a
l
go
rithms
is pr
opo
sed to
ena
bl
e the ro
bot
to r
e
a
c
h the
goa
l p
o
s
ition thr
o
u
gh t
he sh
ortest pat
h.
T
he alg
o
rith
ms
used ar
e fu
zzy logic
and c
e
ll
deco
m
p
o
siti
on
algor
ith
m
s, in
w
h
ich the fu
zzy algor
ith
m
w
h
i
c
h
is an
artificia
l
i
n
telli
ge
nce
alg
o
rith
m is
used
for
robot p
a
th
pla
nni
ng
an
d
cell
deco
m
posi
t
ion a
l
gor
ith
m
i
s
used t
o
cre
a
te
a
map
for the
robot
path,
b
u
t the
mer
ger
of these tw
o a
l
g
o
ri
thms
is still
inc
apa
ble
of fin
d
i
n
g
the shortest pa
th. T
herefore, this
pa
per d
e
sc
ribes a
mo
dific
a
tion of the b
o
th alg
o
rith
ms b
y
addi
ng p
o
ten
t
ial
field
alg
o
rith
m
that is us
ed to
provi
de w
e
i
g
h
t
values
on t
h
e
ma
p i
n
or
der f
o
r the
qua
drot
or to
move
to i
t
s
goa
l pos
itio
n a
nd fin
d
the s
h
ortest path. T
h
e mod
i
fica
tio
n
of the al
gorit
h
m
s h
a
s
sh
ow
n that qua
drotor
is
abl
e to av
oi
d
vario
u
s o
b
stac
les a
nd fi
nd t
h
e shortest
pat
h so th
at the t
i
me re
quir
ed t
o
get to
the
g
oal
positi
on is shor
ter.
Ke
y
w
ords
: Ce
ll Deco
mpos
itio
n, Quadrotor, F
u
zz
y
,
Sh
ortest Path, Modifie
d
Potentia
l F
i
eld
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Quad
roto
r is
an auton
omo
u
s robot that
moves
by u
s
i
ng four
rotors [1] without requiri
ng
an ope
rato
r but a naviga
t
ion system t
hat is ca
pabl
e to dire
ct it to the goal position in
an
unkno
wn are
a
. Quadrotor is one type of UAV us
ed
by SAR tea
m
and the military to perform
missi
on
s su
ch as surveilla
nce, search
and re
scue
v
i
ctims of disa
sters in
the a
r
ea
s. Given the
missi
on,
qua
droto
r
m
u
st f
l
y low in
whi
c
h it
ha
s
a
chall
enge
to
avoid
seve
ral
ob
stacl
e
s.
T
o
overcome th
e pro
b
lem,
several re
se
arche
r
s h
a
ve co
ndu
cted
resea
r
ch
o
n
path pla
n
n
ing
algorith
m
that enable
s
UAV
s
to be able t
o
av
oid ob
sta
c
le
s and mov
e
to the goal positio
n.
Grap
h-se
arch
theory has b
een u
s
ed by Filippis
et al., [2] for UAV
path plan
ning
in a 3
D
environ
ment.
This theo
ry is used to
cre
a
te a
3D
ma
p i
n
an
un
kn
own environm
en
t by dividing
th
e
environ
ment i
n
to small
grid
s. Th
e 3
D
ma
p with
g
r
id
s i
s
a
co
ntour m
ap
with
colo
rs that
re
present
different heig
h
ts above th
e grou
nd. Th
e algorit
hm
use
d
for path plannin
g
are A* and theta*
applie
d to UAV in a 3D environme
n
t. In
variou
s
tests, the theta* algorith
m
has sm
oo
the
r
movement th
an the A*
alg
o
rithm. Th
e d
r
awba
ck of
t
he theta*
alg
o
rithm i
s
that
it is n
o
t able
to
detect dynam
ic ob
stacl
e
s.
In addition
to
the re
se
arch
er me
ntione
d
previo
usly, the theo
ry of
gr
ap
h-se
arch
has al
so
been
devel
o
ped
by G
a
rci
a
et
al [3] fo
r
UAV path
planni
ng i
n
d
ange
rou
s
en
vironme
n
ts.
The
environ
ment i
s
divided into
small g
r
id
s o
n
a 2D m
ap i
n
whi
c
h
seve
ral line
s
a
r
e d
r
awn to co
nn
ect
them. Besi
de
s u
s
in
g 2
D
e
n
vironm
ent
maps, th
e
re
sea
r
che
r
al
so
used a
3
D
e
n
vironm
ent
map
becau
se the UAV works in
that environ
ment. The al
g
o
rithm used i
n
this study is lazy theta wh
ich
is then co
mp
ared
with A* and A*PS (Post-Sm
oothin
g
) algo
rithm
s
. The simulati
on re
sult sh
o
w
s
that lazy
the
t
a algo
rithm
requi
re
s fa
st
er time
tha
n
the
other t
w
o
algo
rithm
s
. However,
this
algorith
m
ha
s merely bee
n applie
d in sta
t
ic but not dynamic e
n
viro
nments.
Grap
h theo
ry
of B-Spline
curve meth
od
use
d
by Elba
nha
wi et al., [4] is a mo
dification
of
B’ezie
r cu
rve
s
used for Car-Like Rob
o
ts path plan
ni
ng. By using Curve
s
B'e
z
i
e
r algo
rithm, the
robot i
s
able
to move to the goal po
sitio
n
and avoid
obsta
cle
s
. Th
e algo
rithm is modified into
a
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 655 – 66
4
656
B-Spline curv
e due to the robot motion that is not sm
ooth. The we
akn
e
ss of this algorithm is t
hat
the robot
can
not avoid mo
ving obsta
cle
s
.
Heu
r
isti
c alg
o
rithm h
a
s b
een u
s
ed
by Yao et al., [5] for robot
path plan
nin
g
in a
n
unkno
wn e
n
vironm
ent. Th
e alg
o
rithm
u
s
ed
is t
he co
mbination
of neural
n
e
two
r
k and
ge
neti
c
algorith
m
s
ap
plied fo
r robot
in a pl
ann
ar
environ
ment
usin
g glo
bal
path pla
nnin
g
method.
Neu
r
al
netwo
rk alg
o
rithm is u
s
ed
to mo
del
a
sta
t
ic envi
r
onm
e
n
t obtain
ed
b
y
usin
g a
glo
bal vie
w
, so t
hat
the starting, g
oal, and ob
stacle po
sition
s of t
he robot
can be vie
w
e
d
, while the genetic alg
o
rit
h
m
is used for
ro
bot path plan
ning. With th
ese al
gorit
h
m
s, the rob
o
t is able to move
to goal po
sition
and avoid
o
b
sta
c
le
s, but
the wea
k
n
e
ss
of this al
gorithm i
s
th
at it is appli
ed only for
static
environ
ment
and be
ca
use of this the rob
o
t cann
ot avoid dynamic o
b
sta
c
le
s.
There
a
r
e h
euri
s
tic algo
ri
thms whi
c
h have
p
r
op
erti
es re
sem
b
lin
g
creatu
r
e
s
su
ch as
bacte
ria
and
bee
s. Foragi
n ba
cteri
a
al
gorithm
used
by Lian
g et
al., [6] is one
of the he
uri
s
ti
c
algorith
m
s h
a
v
ing ch
ara
c
te
ristics li
ke b
a
c
teria i
n
which this alg
o
rith
m can
create
a co
ntour m
a
p
of the enviro
n
ment that is u
s
ed to a
v
oid obs
ta
cle
s
and m
o
ve
to goal po
sition in a st
atic
environ
ment.
Hon
e
y Bee
M
a
ting O
p
timization al
go
rithm u
s
ed
by
Rashmi
et
al., [7] for multi-ro
bot
path plan
ning
has
a ch
ara
c
teri
stic li
ke i
n
se
ct co
lony in
whi
c
h
thi
s
algorith
m
ca
n
make map
s
of
the safest
pa
th and
avoid
ob
stacl
e
s to
move to
go
al po
sition. T
he tra
c
k p
a
th
s
cre
a
ted
usi
n
g
colo
ny bees.
By using this algo
rithm multi-ro
bot can identified
a short pat
h in a dynamic
environ
ment.
Geneti
c
algo
rithms have b
een studi
ed
by Y
ang et al., [8] used for robot path pl
annin
g
algorith
m
co
mbined
with visibility grap
h. Visibilit
y graph alg
o
rithm
is use
d
to create a map o
f
a
n
environ
ment
model
by u
s
i
ng gl
obal
vie
w
,
so th
at th
e ob
sta
c
le
s
positio
n
can
be d
e
termi
n
e
d
,
while
gen
etic algorith
m
is
use
d
to ma
ke rob
o
t path
planni
ng. By usin
g both
of the algo
rith
ms,
robot can avo
i
d obsta
cle
s
a
nd move to the goal po
sitio
n
.
Geneti
c
alg
o
rithms ha
s b
e
en u
s
ed by P
ehlivanogl
u [9] for path pl
annin
g
by co
mbinin
g
Voron
o
i dia
g
ram meth
od.
This
algo
rith
m is
applie
d t
o
a
UAV for
path pla
nnin
g
in an
urban
-l
ike
environ
ment.
In the neigh
borh
ood the
r
e are ob
sta
c
l
e
s in the form of skyscra
per mo
del
s o
r
tall
building
s
. Vo
ronoi
diag
ram
method
is u
s
ed to
create
a ma
p of
an
unkno
wn
envi
r
onm
ent. A
UAV
is set at a certain height, so that with this
algorithm
the UAV will
avoid skyscraper
s and pass
over the build
ings.
The
other h
euri
s
tic algo
rithm such a
s
Ba
cteri
a
l
Potential Fiel
d is a
me
rg
er
of two
algorith
m
s th
at is potenti
a
l field and
bacte
rial
p
o
tential field
algorithm
s
con
d
u
c
ted b
y
a
resea
r
cher
n
a
med Mo
ntiel
et al., [10]. Bacteri
a
l
alg
o
ri
thm is u
s
ed t
o
create a
n
e
n
vironm
ent m
a
p
and a path pl
annin
g
. The wea
k
n
e
ss of the algorithm
is
that it stops at certai
n o
b
sta
c
le
s, so that
it is modified
with the pot
ential field al
gorithm
. With
the modifica
tion of the two algo
rithms,
a
robot is a
b
le to find the sho
r
test path a
n
d
avoid obsta
cl
es in a stati
c
environ
ment.
Heu
r
isti
c algo
rithm wa
s de
veloped by G
hosh et al., [1
1] who appli
e
d Fuzzy algo
rithm for
Micro air vehi
cle path pla
n
n
ing. The alg
o
rithm us
ed
wa
s the devel
opment of fuzzy algorithm
with
quadtree fra
m
ewo
r
k. Qua
d
tree fu
zzy al
gorithm fram
ewo
r
k i
s
the
combi
nation
of fuzzy alg
o
rithm
and graph th
eory. Gra
ph theory is u
s
e
d
to creat
e 2
-
dime
nsi
onal
environ
ment,
while the fuzzy
algorith
m
is u
s
ed for p
a
th planni
ng. The
quadtre
e al
g
o
rithm is u
s
e
d
for path pla
nning in a sta
t
ic
environ
ment
for fixed
win
g
ai
rcraft mo
dels.
Ho
wev
e
r, thi
s
al
go
rithm ha
s
not
bee
n
able
to
determi
ne the
shorte
st path
.
The
other researcher
combined virtual fo
rce fuzzy algorithm
wi
th graph t
heory
as
con
d
u
c
ted by Zhuonin
g
et al., [12]. Graph theory is
u
s
ed to cre
a
te a 2-dime
nsi
o
nal enviro
n
m
ent
map by dividing the map i
n
to small g
r
id
s and the
n
the fuzzy alg
o
ri
thm used to
get to the goal
positio
n and
avoid ob
stacl
e
s is i
dentifi
ed. In or
de
r
to rea
c
h the
goal po
sitio
n
quickly, fuzzy
algorith
m
i
s
combi
ned
wit
h
virtual
force alg
o
rith
m.
The
pro
b
lem
of the
algo
ri
thms i
s
th
at
the
force o
b
taine
d
from virtual
force alg
o
rit
h
m is co
nsi
d
erably g
r
eat that results in
frequent fail
s in
avoiding ob
st
acle
s.
Fuzzy algo
rit
h
m ha
s be
en
use
d
by Wa
ng et
al., [13] for path pl
a
nning in
an u
n
kn
own
environ
ment.
The algo
rith
m is com
b
in
ed with exa
c
t cell de
comp
osition g
r
ap
h
method use
d
to
cre
a
te a ma
p
by dividing the tra
c
ks into
equal
si
ze t
o
avoid ob
sta
c
le
s and
rea
c
h the goal a
n
d
the fuzzy alg
o
rithm i
s
u
s
e
d
for path
pla
nning. To
rea
c
h
the goal
p
o
sition,
the algorithm doe
s not
s
e
ac
rh for the s
hortes
t
path.
Several al
go
rithms p
r
eviou
s
ly mentio
ne
d hav
e
not b
een a
b
le to i
dentify the shorte
st
path and ca
nnot be appl
ied in a dynamic environ
ment. Hen
c
e
,
this paper
pre
s
ent
s fuzzy
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Quad
roto
r Path Planning B
a
se
d On Mod
i
fied Fu
zzy
Cell De
com
position Algorith
m
(Iswanto
)
657
algorith
m
s
combine
d
with cell
de
comp
osition
meth
od an
d mo
di
fied by ad
din
g
potential
field
algorith
m
in o
r
de
r to rea
c
h
the goal po
sit
i
on more rapi
dly and avoid
dynamic ob
stacle
s.
2. Non Linea
r
Quadro
tor
Modeling
Quad
roto
r is an unma
nne
d aircraft with
rotary wing t
y
pe using 4 rotors at ea
ch
end of
the frame tha
t
moves with
thrust as sh
own in Fi
gu
re 1 which is a model of Mahony et al [14]
pre
s
ente
d
in t
h
is p
ape
r. In the figure it is
see
n
that the
quad
roto
r mo
ves by u
s
ing
the four
rotors
whi
c
h rotate i
n
opp
osite
direction i.e. two roto
rs
rotat
e
in a
clo
c
kwi
s
e di
re
ction
a
nd the two ot
her
rotors rotates counte
r
cl
ockwise.
Figure 1. Qua
d
roto
r modelli
ng
Quad
roto
r system
i
s
a
n
o
n
-
linea
r syste
m
that
can
b
e
controll
ed
b
y
six out
of th
e twelve
states that re
gulate the attitude
of quad
rotor
system
that is. The
first six state
s
involves Euler
angle
s
con
s
isting of roll, pitch, yaw a
ngle
s
den
oted a
s
T
2
and an
gula
r
spe
ed o
n
three o
r
thog
o
nal axes of t
he body of quadrotor d
e
n
o
ted as
T
r
q
p
4
. Whil
e six other
states
are
th
ree axis posi
t
ions denote
d
a
s
T
m
l
k
1
and th
ree li
nea
r vel
o
city of th
e
cente
r
of m
a
ss of
the
qua
drot
or
asso
ciated
with fixed
referen
c
e
fra
m
e d
enoted
as
T
m
l
k
3
. So that the
equatio
ns of
non-li
nea
r mo
del of the qua
droto
r
are a
s
follows:
k
x
l
y
m
z
(1)
r
t
c
q
t
s
p
r
s
q
c
r
c
c
q
c
s
(2)
s
s
c
s
c
T
m
k
1
c
s
s
s
c
T
m
l
1
c
c
T
m
g
m
1
(3)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 655 – 66
4
658
qr
db
p
x
y
z
x
2
4
2
1
2
3
2
2
pr
y
db
q
y
z
x
2
2
2
1
2
4
2
3
pq
k
r
z
x
y
z
2
3
2
1
2
4
2
2
(4)
3. Rese
arch
Metho
d
Path plannin
g
is an algo
ri
thm that is neede
d
by a quadrotor in o
r
de
r to reach
its goal
positio
n an
d
avoid o
b
sta
c
l
e
s. T
here a
r
e two
meth
o
d
s
of path
pl
annin
g
that i
s
gl
obal
and
local
path pla
nning
s. Lo
cal pat
h
plannin
g
is
planni
ng t
hat
use
s
sen
s
o
r
s attached to
the body of
the
robot
so
that
the ro
bot id
e
n
tifies me
rely
the o
b
st
a
c
le
s pl
aced i
n
front of it, whil
e the
glob
al
path
planni
ng use
s
sen
s
o
r
s
to see
glo
bally. This
pa
per
p
r
ese
n
ts a re
se
arch
m
e
thod
for
the
stu
d
y of
global p
a
th pl
annin
g
. The rese
arch m
e
th
ods
used in t
h
is p
ape
r are
map cre
a
tion
, path planni
n
g
,
quad
roto
r ap
plicatio
n and
potential valu
e provisi
on to
the map.
3.1. Map Cre
a
tion
Cell de
com
p
osition i
s
on
e of the methods in g
r
a
ph theory u
s
ed to creat
e two-
dimen
s
ion
a
l
maps for
rob
o
t path pla
nni
ng. Thi
s
meth
od divide
s th
e map i
n
to a
several g
r
id
s
that
serve
s
to m
a
ke the
rob
o
t path. The
r
e a
r
e two
me
tho
d
s that i
s
ap
proximate
cel
l
decompo
siti
on
and
exact
cel
l
de
comp
ositi
on. App
r
oxim
ate cell de
co
mpositio
n i
s
use
d
by Val
e
nte et al., [15
]
for
a quad
rotor p
a
th plannin
g
coverage. An
environme
n
t is divided into
several e
qua
l grids by u
s
i
n
g
these m
e
thod
s. Then
a pat
h is
cre
a
ted o
n
the gri
d
s l
o
cated i
n
the
map. Path pl
annin
g
cove
rage
algorith
m
is
use
d
in thi
s
study, so tha
t
t
he qua
drot
or move
s to
the goal
po
si
tion and
avoi
d
obsta
cle
s
wit
hout se
eki
ng
the sho
r
test p
a
th.
Path plannin
g
coverage
algorith
m
ha
s bee
n us
ed
by Galcera
n
& Carrera
s
[16] for
different path
planni
ng coverag
e
to cre
a
t
e a map in
a
n
un
kno
w
n e
n
vironm
ent b
y
using
exact
cell
decompo
sitio
n
. Several g
r
ids ma
de in
the map
by u
s
ing thi
s
al
go
rithm have
e
qual
size. Pa
th
planni
ng coverag
e
algo
rit
h
m is u
s
ed fo
r the grid divi
sion. It is app
lied for floor
clea
ning robo
t in
whi
c
h the ro
b
o
t achieve
s
the goal a
nd a
v
oid obs
ta
cle
s
witho
u
t see
k
ing the
sho
r
test path.
Figure 2. Environm
ent mod
e
l
Approximate
cell de
co
mpo
s
ition alg
o
rith
m is pr
esente
d
in this pa
pe
r as
sho
w
n i
n
figure
2. It sho
w
s that there a
r
e
initial
potition
of (Xr, Yr), st
atic ob
sta
c
le
s positio
n of
(Xo, Yo), and
the
goal p
o
sitio
n
of (Xg, Yg)
in an u
n
kno
w
n e
n
viron
m
ent by viewin
g glob
ally. Using th
e po
si
tion
data, an
envi
r
onm
ent ma
p
is o
b
tained.
The
sou
r
ce
code
sho
w
n i
n
Figure 3 i
s
u
s
ed to
create
a
map and
split
it into severa
l grids.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Quad
roto
r Path Planning B
a
se
d On Mod
i
fied Fu
zzy
Cell De
com
position Algorith
m
(Iswanto
)
659
Figure 3
sho
w
s th
at
R
R
Y
X
,
is th
e po
sition of
the rob
o
t,
O
O
Y
X
,
is the po
sition
of
obsta
cle 1, a
nd
G
G
Y
X
,
is the po
sition of goal.
G
R
is the formul
a of the dista
n
ce b
e
twe
en
the
positio
n of th
e ro
bot an
d t
he go
al an
d
O
R
is the fo
rmul
a
of the di
stan
ce b
e
twe
en t
he po
sition
o
f
the robot an
d
the obsta
cle.
Then
G
R
and
O
R
are summe
d to an enviro
n
m
ent matrix value.
Figure 3. Matrix environm
e
n
t of cell decompo
sition
3.2. Fuzzy
Path Planning
Basic b
ehavi
o
r of robot p
a
th plannin
g
in dec
ompo
si
tion cell algo
rithm is go-to
-goal. In
this behavio
r,
the robot sea
r
ch
es fo
r the smalle
st gr
id
to move by grid following to
get to the goal
positio
n. Cell
decom
po
sition algo
rithm
is a path pla
nning al
gorit
hm in
2
environment while
positio
n of q
u
adroto
r
i
s
in
2
environ
ment.
Hen
c
e
this
p
aper p
r
opo
se
d qu
adrotor
p
a
th pla
nning
with fuzzy algorithm for
2
environ
menta
l
position by providin
g a consta
nt value
at a certain
height. In this study, the quadrotor
i
s
controlle
d with
refere
nce values of gri
d
spaci
ng obtain
e
d
from fuzzy algorithm.
This algo
rith
m which
i
s
a
n
a
r
tificial int
e
lligen
ce
alg
o
rithm
used f
o
r
rob
o
t navi
gation i
n
unkno
wn env
ironm
ent is d
i
scovere
d
by
zad
eh
[15]
and u
s
ed
by Lee [16]. F
u
zzy input is an
ultrasoni
c se
nso
r
attached
to the front body of
the ro
bot with angl
e betwe
en se
nso
r
s i
s
30
° an
d
there a
r
e 12
sen
s
o
r
s. Fu
zzy algorith
m
use
d
ha
s a
rule ba
se by compa
r
ing det
ection di
stan
ce of
obje
c
ts to
det
ection
an
gle
of obje
c
t. Th
e fuzzy
al
gori
t
hm ha
s
su
ccessfully pa
ssed the
ob
sta
c
le
toward goal p
o
sition.
The othe
r re
searche
r
s su
ch as Kala et
al., [
17] comb
ining fuzzy al
gorithm
with A* used
for ro
bot path
planni
ng. A*
algorith
m
is
u
s
ed fo
r path
p
l
annin
g
on
co
lored
map
s
created by u
s
in
g
grap
h algo
rithm, while fuzzy algorithm i
s
used fo
r na
vigating rob
o
t. With the merge
r
of the two
algorith
m
s, th
e robot
can a
v
oid obsta
cle
s
and rea
c
h its goal p
o
sitio
n
.
Figure 4. Fuzzy path plan
n
i
ng de
sign
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 655 – 66
4
660
Figure 5. Set of fuzzy input
and output
Figure 6. Fuzzy path plan
n
i
ng rule b
a
se
The fu
zzy i
n
put of g
r
id v
a
lue
s
is obt
ained f
r
om
cell de
com
p
o
s
ition o
n
gl
o
bal path
planni
ng
whil
e the mem
b
e
r
set fuzzy ou
tput is motio
n
dire
ction
of quad
roto
r a
s
sho
w
n i
n
figu
re
4. The
path
p
l
annin
g
p
r
op
o
s
ed
in thi
s
study is for th
e
rob
o
t in
2
R
po
sition while q
u
adroto
r
i
s
a
robot that i
s
in
3
R
positio
n. Therefore, th
e quad
roto
r
will wo
rk in
2
R
position in which the
altitude is set up.
The fu
zzy m
e
thod u
s
e
d
i
n
this
pap
e
has eight i
n
p
u
ts
con
s
i
s
tin
g
of three m
e
mbe
r
s ie
small, me
diu
m
, and la
rge
rangi
ng from
0 to 300
a
s
shown in Fi
gu
re 5. T
here i
s
on
e outp
u
t
of
fuzzy
and
it
has eig
h
t m
e
mbe
r
s de
riv
ed from th
eir pla
c
e
s
nam
ely obliq
ue f
r
ont ri
ght, fro
n
t,
oblique front l
e
ft, left, oblique rear l
e
ft, rear, obli
que
rear
right, right and. The fuzzy path pl
anning
rule ba
se i
s
shown in Figu
re 6.
3.3. Potentia
l Field
The m
e
rging
of both
algo
rit
h
ms en
able
s
quad
roto
r to
avoid o
b
sta
c
l
e
s
and
go
to t
he g
oal
positio
n, but
the qu
ad
rotor ha
s
unide
ntified the
s
hort
e
st p
a
th. Th
e alg
o
rithm
o
f
this
pap
er i
s
modified
by a
dding
potenti
a
l field al
go
rithm u
s
ed
to gi
ve a weight v
a
lue fo
r e
a
ch
grid
cell
form
ed
from de
comp
osition. Poten
t
ial field algorithm is
an alg
o
rithm for ro
b
o
t navigation that is used b
y
Park
et al., [20] usi
ng m
agneti
c
theo
ry that is
attraction
and
repul
sion. In
robot
navigat
ion,
attractive pot
ential is the force gen
erat
ed by the
goal potition of robot in
which the equatio
n is as
follows
:
2
2
1
)
(
d
a
A
x
x
k
x
U
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Quad
roto
r Path Planning B
a
se
d On Mod
i
fied Fu
zzy
Cell De
com
position Algorith
m
(Iswanto
)
661
2
2
1
)
(
d
a
A
y
y
k
y
U
, (5)
Whe
r
e
a
k
is th
e
potential fiel
d consta
nt. T
he value
of
d
d
y
x
,
sho
w
s the
go
al po
sition
of the
robot a
nd
y
x
,
sh
ows the
coo
r
dinate p
o
sitio
n
of the ro
bot
. While p
o
ten
t
ial repul
sio
n
is a force
gene
rated by
obsta
cle
s
in robot enviro
n
m
ent issho
w
n
in the followi
ng equ
ation:
2
1
1
2
1
)
(
O
r
R
x
x
k
x
U
2
1
1
2
1
)
(
O
r
R
y
y
k
y
U
, (6)
Whe
r
e
r
k
is the
repul
sive
co
nstant. The v
a
lue of
o
o
y
x
,
sho
w
s t
he o
b
st
a
c
l
e
po
sit
i
on.
I
n
t
h
is
pape
r the p
o
tential field in
attractive an
d
repul
sive e
q
uation
s
is m
o
dified to be
u
s
ed to
provid
e
the value of the grid o
n
the
cell de
comp
os
ition. The
modificatio
n
of the equatio
n is:
;
10
10
2
2
G
R
G
G
a
A
X
X
Y
Y
k
U
(7)
;
10
10
2
2
S
S
R
S
R
r
R
X
X
Y
Y
k
U
(8)
;
10
10
2
2
D
D
R
D
R
r
R
X
X
Y
Y
k
U
(9)
Whi
c
h
D
R
U
is the
potential field
force for dyn
a
mic o
b
sta
c
l
e
s,
S
R
U
is the pot
ential field force f
o
r
static obst
a
cl
es,
an
d
A
U
is t
he pote
n
tial f
i
eld atractio
n. While
G
G
Y
X
,
is th
e goal
po
siti
on,
S
S
Y
X
,
is the static o
b
sta
c
le po
siti
on, and
D
D
Y
X
,
is the dynamic o
b
s
tacl
e po
sitio
n
.
3.4. Modifed
Map Cre
a
tio
n
This pap
er u
s
e
s
a
dyna
mic e
n
viron
m
en
t whe
r
e
the
e
n
vironm
ent h
a
s
a m
o
ving
obsta
cl
e
as
sho
w
n i
n
Figure 6. In the figure
G
R
is the dista
n
ce b
e
twen th
e ro
bot and th
e
goal p
o
sitio
n
,
D
O
R
is the dista
n
ce betwe
n the
robot
an
d the dynamic
ob
stacl
e
, and
S
O
R
is a dista
n
ce
betwe
n
the robot
and
the
static
ob
stacl
e
. By u
s
i
ng a
cell d
e
compo
s
ition, t
he e
n
viron
m
e
n
t is
divided
i
n
to
grid
s with sou
r
ce
cod
e
as i
n
Figure 7.
Figure 8
sho
w
s th
at cell
d
e
com
p
o
s
ition
algorit
h
m
is
modified to u
pdate the i
n
formatio
n
data of the e
n
vironm
ent that is use
d
to detect mo
vin
g
obsta
cle
s
so that the algorithm is pl
ace
d
in the
main
source
code. I
n
this figu
re
K
is the
value
f
o
r th
e
size of
the
grid,
R
R
Y
X
,
is
th
e
positio
n of th
e robot,
G
G
Y
X
,
is t
he
robot
go
al po
sition,
S
S
O
O
Y
X
,
is
the po
sition
of stati
c
obsta
cle
s
, an
d
D
D
O
O
Y
X
,
is the po
sition of dynami
c
ob
st
acl
e
s. Having creat
ed
an
envi
r
on
ment
map usi
ng ce
ll decom
po
sition, fuzzy pat
h planni
ng al
gorithm i
s
performed.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 655 – 66
4
662
Figure 7. Dyn
a
mic envi
r
on
ment model
Figure 8. Modified cell d
e
compo
s
ition
The
novelty o
f
the alg
o
rith
m in thi
s
stud
y other
than
u
pdated
inform
ation i
s
to
mo
dify cell
decompo
sitio
n
alg
o
rithm
to p
r
ovide
weight valu
es
to ea
ch
gri
d
with
modifie
d
pote
n
tial fi
eld
algorith
m
. From Equation
(7), (8), and
(9) the value for ea
ch g
r
id o
b
tained i
s
:
D
S
R
R
A
U
U
U
F
(10
)
Equation
(1
0) is
used to
provide value
s
to
grid
s by
m
odifying
the
source co
de a
s
sho
w
n
in Figure 9.
Figure 9. Matrix environm
e
n
t of modified cell de
comp
osition
Matrix enviro
n
ment of mo
dified cell d
e
c
omp
o
sitio
n
is sh
own in Figure 9
in which
the
total force of
potential field is use
d
for weight
value
s
of colum
n
s and ro
ws in
it.
If
the sum of
attraction a
n
d
repul
sion i
s
consi
derably large, a limit value is give
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Quad
roto
r Path Planning B
a
se
d On Mod
i
fied Fu
zzy
Cell De
com
position Algorith
m
(Iswanto
)
663
3. Results a
nd Analy
s
is
The exp
e
rim
ents i
n
thi
s
pap
er i
s
condu
cted
by
usi
ng M
a
tla
b
software t
o
mod
e
l
environ
ment
and qu
adroto
r
and al
so
si
mulate the
pe
rforma
nce of quad
roto
r in reaching the g
oal
positio
n and
avoiding stati
c
ob
stacle
s in
an unkn
o
wn environ
ment. There are two experime
n
ts in
this pa
per to test the
modified al
g
o
rithm
s
.
The
first expe
ri
ment u
s
e
s
static and
dynami
c
obsta
cle
s
a
n
d
the
se
cond
experim
ent u
s
e
s
stati
c
o
b
stacle
s. The
e
x
perime
n
ts
compa
r
e m
odif
i
ed
Fuzzy Cell
De
comp
ositio
n Artificial P
o
tentia
l Field (FCDAPF) t
o
Fuzzy Cell Decompositi
o
n
(FC
D
) al
go
rithm.
The first expe
riment te
sts t
he FCDAPF
algor
ith
m
in a
n
un
kno
w
n e
n
vironm
ent containe
d
static a
nd dy
namic
ob
stacles a
s
sho
w
n
in Figure 8.
The figure sh
ows that t
he
static o
b
sta
c
l
e
is
locate
d at
the
po
sition
of (5
,7) a
nd th
e d
y
namic ob
sta
c
le i
s
located
at the
po
sitio
n
of
(9,1
) a
n
d
it
moves to
po
sition
(1,7
). T
he e
n
vironm
ent teste
d
with FCD
algo
rithm is sho
w
n
in Fig
u
re 10
(a
)
and
FCDAPF
in Fi
gu
re
10(b). In
Figu
re
10(a
)
, it
ca
n b
e
seen
that th
e g
r
ee
n q
uad
rotor will
colli
de
with the yello
w on
e at the
positio
n of (5
, 5), wh
il
e Fi
gure
10
(b
) sh
ows that the
gree
n q
uad
ro
tor
will avoid the
moving ob
sta
c
le of yellow
quad
roto
r at the po
sition of
(4.5, 4.5).
Figure 10. Experim
ent with
FCD a
nd FCDAPF algo
rithms
In this exp
e
ri
ment the
rob
o
t is pl
aced
a
t
t
he initial p
o
s
ition of
(1
0,1
0
) m
o
ving to
the goal
positio
n of
(9
0, 100
) in
the
enviro
n
me
nt having
four
circula
r
ob
sta
c
le
s lo
cate
d
at the p
o
sitio
n
of
(3, 3),
(3,6
), (6, 2.5), a
nd
(6, 7) a
s
sho
w
n in Fig
u
re
1
1
. The figu
re
is si
mulated
by two meth
o
d
s
namely fu
zzy cell
deco
m
positio
n m
e
thod i
ndica
ted by the
gre
en li
ne
and
fuzzy
cell
decompo
sitio
n
that has be
en modified
with the pote
n
tial field indicated by the blue line. It shows
that by using
modified fuzzy cell de
co
mpositio
n
alg
o
rithm, the quadrotor rapi
dly moves to the
goal po
sition
and sta
b
ly flies be
cau
s
e it moves in ea
ch grid.
Figure 11. Experim
ent with
FCD a
nd FCDAPF algo
rithms
0
10
20
30
40
50
60
70
80
90
10
0
0
10
20
30
40
50
60
70
80
90
100
Me
t
e
r
Me
t
e
r
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 655 – 66
4
664
4. Conclusio
n
The p
ape
r p
r
ese
n
ts
a ne
w algo
rithm u
s
i
ng the
merge
r
of
cell
de
co
mpositio
n a
n
d
fuzzy
algorith
m
s. F
u
zzy algo
rith
ms i
s
typical
l
y used
for
path pla
nnin
g
whi
c
h
serv
es to
dete
c
t a
n
obsta
cle
s
located in front of the robot by using se
n
s
ors. In this study the alg
o
rithm is u
s
e
d
to
detect
ob
stacles that
are re
ad fro
m
the
cell de
com
p
o
s
i
t
ion algo
rithm
.
By using th
e
algo
rithm, th
e
quad
roto
r i
s
able to
move
to the g
oal
po
sition
and
av
oid o
b
sta
c
le
s but it
requi
re
s a
lon
g
p
r
o
c
ess
to re
ach the
goal. S
o
th
at an
additio
nal al
gorith
m
to resolve t
he i
s
sue i
s
need
ed.
With
the
merg
er of the three al
g
o
rithm
s
, the quadrotor i
s
able to a
v
oid obsta
cl
es in a dyn
a
mic
environ
ment
and move
s to the goal po
si
tion more q
u
i
ckly.
Referen
ces
[1]
R Abb
a
s, Q W
u
. Improved
Lea
der F
o
llo
w
e
r F
o
rmati
on
control
l
er for
multip
le Qu
a
d
rotors b
a
se
d
AFSA.
TELKOMNIKA.
2015; 13(1): 85-
92.
[2]
L Filip
pis, G Gugli
e
ri, F Quag
liotti. Path Pl
a
nni
ng Strateg
i
e
s
for UAVS in 3D Envir
onme
n
ts.
J. Intell.
Robot. Syst.
2
011; 65(
1-4): 2
47-2
64.
[3]
M Garcia, A. Viguri
a
, A Ollero
. D
y
nam
ic gra
p
h
-searc
h
alg
o
ri
thm for globa
l path pl
an
nin
g
i
n
prese
n
ce
of hazard
ous
w
eather.
J. Intell
. Robot. Syst.
T
heory App
l
.
2013; 69(
1-4): 2
85-2
95.
[4]
M Elbanha
w
i
, M Simic, RN Jaza
r. Conti
n
u
o
u
s Path Smoot
hin
g
for Car-L
i
k
e Rob
o
ts Usi
ng B-Spl
i
n
e
Curves.
J. Intell. Robot. Syst.
201
5; 1(8): 1-3
4
.
[5]
P Yao,
H W
ang, Z
Su.
U
AV feasi
b
l
e
p
a
th p
l
a
nni
ng
base
d
on
dist
urbe
d fl
uid
a
n
d
traj
ector
y
prop
agati
on.
C
h
in
ese J. Aero
naut
. 20
15; 28(
4): 1163-
11
77.
[6]
X
Li
ang,
L
Li,
J W
u
, H
Ch
en. Mo
bil
e
r
o
b
o
t path
p
l
an
ni
ng
bas
ed
on
ada
ptive
bacte
rial f
o
rag
i
n
g
algorithm.
J. Cent. South Un
i
v
.
2013; 20(
12)
: 3391-3
4
0
0
.
[7]
RS Rashmi, R
Prat
y
u
sh
a, T
H
Md, S Sujata, M
Sharmilla, P Rajara
jes
w
a
r
i, C Sanja
y
, N Nag
w
a
n
i, D
Lop
amudr
a, C
Ramb
abu,
et a
l
. Navi
gati
ona
l
Path Pl
ann
in
g of
Multi-R
obot usin
g
H
one
y B
ee
Mati
ng
Optimization Algorithm (HBMO).
Int
.
J.
Comput. Appl.
20
11
; 27(11): 1-8.
[8]
X Ya
n. An Improved R
o
b
o
t Path Planning Algorithm.
TELKOMNIKA.
2012; 10(4): 62
9-6
3
6
.
[9]
YV Peh
liva
n
o
g
l
u. A n
e
w
v
i
br
ation
a
l
gen
etic
alg
o
rithm
en
h
ance
d
w
i
t
h
a
Voron
o
i
dia
g
ra
m for path
pla
nni
ng of aut
onom
ous UAV.
Aerosp. Sci. Technol
. 20
12; 16(1): 47-
55.
[10]
O Montiel, U
Orozco-Ros
as, R Sep
ú
lve
da.
Path
pl
an
nin
g
for mobi
le r
obot
s usin
g Bacteri
a
l Pote
nti
a
l
F
i
eld for avo
i
di
ng static an
d d
y
n
a
mic o
b
stacl
e
s.
Expert Syst. Appl.
201
5; 42(12): 51
77-
51
91.
[11]
S Ghosh, A Haid
er, M Sinha
. Micro air veh
i
cle p
a
th pl
an
n
i
ng i
n
fuzz
y
qu
adtree frame
w
ork.
J. Inst.
Eng. Aerosp. E
ng. J.
2010; 91
: 10-15.
[12]
D Z
h
u
oni
ng, Z
Ruli
n, C
Z
o
n
g
ji
, Z
Rui. St
ud
y
on
UAV P
a
th P
l
an
nin
g
A
ppro
a
c
h Bas
e
d
on
F
u
zz
y Virtu
a
l
Fo
rce
.
Chin
ese
J. Aeronaut.
2
010; 23(
3): 341
-350.
[13]
Y W
ang, T
Wei, X Qu. St
u
d
y
of multi-o
b
j
e
ctive fuzz
y
o
p
timizati
on for
path pl
ann
in
g
.
Chines
e J.
Aeron
aut.
201
2; 25(1): 51-5
6
.
[14]
R Mahony
, V
Kumar, P
Corke. Multirotor
Aeri
a
l
V
ehic
l
es: Mod
e
li
ng,
Estimation,
an
d Co
ntrol
of
Quadrotor.
Robot. Autom
.
Mag. IEEE.
2012; 19(3): 20-
32.
[15]
J Val
ente,
D S
anz, J
Del
C
e
r
r
o, A Barr
ie
nto
s
, MÁ
de
F
r
ut
os. Ne
ar-optim
al c
o
vera
ge
tra
j
ectori
es for
imag
e mosaic
i
ng usi
ng a mi
ni
quad-r
o
tor ove
r
irregu
lar-sh
a
p
ed fiel
ds.
Precis. Agric.
2013;
14(1): 11
5-
132.
[16]
E Galceran, M Carreras. A surve
y
on cover
age p
a
th pla
n
n
i
ng for robotics
.
Rob. Auton. Syst
. 2013;
61(1
2
): 125
8-1
276.
[17]
LA Z
ade
h. Outl
ine
of a N
e
w
A
ppro
a
ch to th
e
Anal
ys
is of Co
mple
x S
y
stems
and
Decis
i
o
n
Processes.
Syst. Man Cybern. IEEE Trans.
1973; 3(1): 2
8
-44.
[18]
T
Lee. F
u
zzy
Motion Pl
an
nin
g
of
Mobi
le R
obots i
n
Unk
n
o
w
n E
n
viro
nm
ents.
J. Intell. Robot. Syst
.
200
3; 37: 177-
191.
[19]
R Kal
a
, A Sh
ukla, R
T
i
w
a
r
i
. Fusion
of pr
o
bab
ilistic
A
* alg
o
rithm and
fuzz
y
infere
nc
e
s
y
stem
for
robotic p
a
th pl
ann
ing.
Artif. Intell. Rev.
201
0;
33(4): 307-
32
7.
[20]
M Park, J Jeon, M Lee.
Obst
acle
avo
i
d
ance
for
mo
bil
e
ro
b
o
ts usi
ng
artific
i
al
pote
n
tia
l
fie
l
d a
ppro
a
ch
w
i
th simul
a
ted
anne
ali
ng.
In IEEE Internationa
l S
y
m
posi
u
m on
Industria
l Electron
ics, 200
1. 200
1
:
153
0-15
35.
Evaluation Warning : The document was created with Spire.PDF for Python.