Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
Vol.
5, No. 6, Decem
ber
2015, pp. 1372~
1
380
I
S
SN
: 208
8-8
7
0
8
1
372
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Optimized PID Controller with
Bacterial Foraging Algorithm
Seiyed Mohammad
Mirz
ae
i*,
Mohammad H
o
ssein
Moattar
**
* Is
lam
i
c
Azad
Univers
i
t
y
,
F
e
rd
ows
Branch,
F
e
r
dows
,
Iran
** Departmen
t
o
f
Software Engineering
,
Mashhad
Branch
, Is
l
a
m
i
c Azad
Univers
i
t
y
,
M
a
s
hhad
,
Ir
an
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
May 8, 2015
Rev
i
sed
Ju
l 27
,
20
15
Accepte
d Aug 9, 2015
Fish robot precis
i
on depends on a variet
y
of factor
s including th
e p
r
ecision of
m
o
tion sensors,
m
obilit
y
of l
i
nk
s, elasti
cit
y
of f
i
sh robot actuat
o
rs sy
stem
,
and the precision of controllers. Am
ong
these factors, precision and
efficiency
of con
t
rollers play
a key
role
in fish robot precision
. In the presen
t
paper, a robot fish has been
desi
gned with
d
y
n
a
mics and
swimming
mechanism of a
real f
i
sh. Accord
ing to eq
u
a
tions of motion, th
is fish robot is
designed with 3
hinged links. Subsequent
ly
, its
control s
y
stem
was defined
based on the same equations. I
n
this pa
per, an
approach is suggested to
control
fish robo
t tr
aje
c
tor
y
using
optim
ized
PID controller throug
h
Bacterial
Foraging a
l
gorit
hm
, so as to
adjust
the g
a
in
s. Then
,
this
c
ontrolle
r is
com
p
ared to th
e powerful F
u
z
z
y
control
l
er
an
d optim
ized P
I
D controlle
r
through PSO algorithm when apply
i
ng step an
d sine inputs. The resear
ch
findings revealed that optimized
PI
D controller
through Bacterial Foragin
g
Algorithm had
better p
e
rfor
m
ance th
an other appro
ach
es in terms of
decreasing
of th
e
settling time, re
duction of
the maxi
mum overshoot and
desired stead
y
state error in r
e
spons
e to step
input. Eff
i
cien
cy
of
the
suggested metho
d
has been
analyzed b
y
MA
TLA
B software.
Keyword:
Bacterial foraging al
gorithm
Fish
r
obo
t
Fuzzy c
ont
roll
er
PID con
t
ro
ller
PSO algorithm
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Seiyed Mohammad Mirzaei,
Islamic Azad
Uni
v
ersity,
Ferdows B
r
anc
h
,
Ferdows
,
B
o
j
n
ou
rd
, No
rt
h Kh
or
asan
, Ir
an.
Em
a
il: M.
mirzaie9
0@yaho
o.co
m
1.
INTRODUCTION
Ro
bo
ts are th
e m
a
n
i
festatio
n
o
f
th
e m
o
st ex
citin
g
a
d
vanc
ed technology in
prese
n
t e
r
a
,
especially
t
hose t
h
at
are
i
n
spi
r
e
d
by
t
h
e surr
o
u
n
d
i
ng
nat
u
re. C
r
a
w
l
y
, ru
nni
n
g
, fl
y
i
ng a
nd s
w
i
m
mi
ng r
o
b
o
t
s
at
t
r
act
t
h
e
at
t
e
nt
i
on o
f
al
m
o
st
any
one. I
n
spi
t
e
o
f
t
ech
nol
ogi
cal
ad
va
nces i
n
t
h
e c
o
nst
r
uct
i
on
o
f
m
obi
l
e
rob
o
t
s
suc
h
a
s
hum
anoid, res
c
ue, s
o
ccer
pla
y
er, wa
rrior, a
nd sc
out robo
t
s
, etc., not enough re
searc
h
es
have
bee
n
done on
u
n
d
e
rwater robo
ts
p
a
rticu
l
arly
on
fish-lik
e rob
o
t
s yet [1
].
Usu
a
lly fo
r linear and
no
n-lin
ear
syst
e
m
s, classic cont
roll
ers are
us
ed
due to their sim
p
le structure
an
d
po
werfu
l
p
e
rform
a
n
ce in
in
du
strial con
t
ro
lling
pr
o
c
edu
r
es. Sev
e
ral qu
an
titativ
e m
e
t
h
od
s
h
a
v
e
b
e
en
u
s
ed
suc
h
as Fuzzy
Logi
cal
C
o
nt
rol
w
h
i
c
h i
s
a cont
r
o
l
l
i
ng
method base
d on the fuzzy
logic, as well as PID
cont
rol
l
e
r
[
5
]
.
PID
c
ont
r
o
l
l
e
r
i
s
a m
echani
s
m
wi
t
h
a cl
ose
d
-l
oo
p
feed
bac
k
.
Desi
gni
ng
s
u
ch
a c
o
nt
rol
r
e
qui
res
t
uni
n
g
o
f
t
h
ree
val
u
es, p
r
op
o
r
t
i
onal
gai
n
(K
p
), Inte
gral gai
n
(K
i
) an
d de
r
i
vat
i
v
e gai
n
(
K
d
). T
h
e goal of the
co
n
t
ro
ller is to
m
i
n
i
mize sig
n
a
l error, rise ti
me, o
v
e
rshoo
t an
d
settling
time th
ro
ugh
tun
i
ng
th
e inpu
t in
th
e
co
n
t
ro
l
pro
cess. Rise tim
e is
th
e tim
e it ta
k
e
s
for t
h
e syste
m
respo
n
s
e to
step
inpu
t to reach
1
0
-9
0
%
of it
s
stable val
u
e.
Settling ti
m
e
refers t
o
the
ti
me it takes
for the system
step response
t
o
reach a ce
rtain
range
aroun
d
its fi
n
a
l v
a
lu
e for th
e first ti
m
e
, also
to
rem
a
in
co
n
s
tan
t
. Th
is ran
g
e
is
often
stated
as an
ab
so
l
u
te
perce
n
t
a
ge
o
f
t
h
e a
b
s
o
l
u
t
e
val
u
e
(2
-5
%)
[
17]
.
A
lo
t
of
r
e
search
es
h
a
v
e
inv
e
stig
ated
f
i
sh
ro
bo
ts. In
1
994, th
e
f
i
r
s
t
r
obotic f
i
sh
n
a
m
e
d
Rob
o
T
un
a
was produced at
MIT [6].
It was
s
u
ccess
f
ully devel
ope
d i
n
to an 8-link R
o
boT
u
na
whic
h m
a
y be the
fi
rst free-
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Op
timized
PID Con
t
ro
ller with
Ba
cteria
l Fora
g
i
ng
Algo
rit
h
m (S
eiyed Moh
a
mmad
Mirzaei
)
1
373
swi
m
m
i
ng ro
b
o
t
i
c
fi
sh i
n
t
h
e
wo
rl
d [
7
]
.
I
n
spi
r
e
d
by
t
h
i
s
st
udy
, t
h
e
Dra
p
er La
b
o
rat
o
ry
devel
ope
d u
n
d
ersea
vehi
cl
e w
h
i
c
h
i
s
nam
e
d VC
UU
V. Si
nce t
h
i
s
ve
hi
cl
e co
ul
d a
voi
d o
b
st
acl
es, and i
t
i
s
capabl
e
o
f
u
p
-
d
o
w
n
m
o
ti
on;
i
t
i
s
t
h
e
m
o
st
kno
wn
ro
b
o
t
i
c
fi
sh [7]
-
[
9
]
.
Aft
e
r i
t
s
d
e
vel
o
pm
ent
,
researche
r
s de
ve
l
ope
d m
a
ny
kinds
of
robo
tic fish
es. As an
instan
ce, in
Na
goya
Uni
v
ersity, Ja
pan, a sm
all robot fis
h
was
designe
d
usi
n
g ICPF
act
uat
o
rs [
8
]
-
[
10]
. A
ki
nd o
f
ro
b
o
t
i
c
fi
sh na
m
e
l
y
G9
(9.
G
e
n
erat
i
o
n) was devel
ope
d
i
n
E
ssex Uni
v
ersi
t
y
whi
c
h
h
a
d th
e
b
e
st swi
mmin
g
ab
ility and
n
o
w, it is
ex
h
i
b
itin
g
i
n
Lo
ndo
n Coun
ty Hal Aqu
a
ri
u
m
[8
].
A m
a
th
e
m
a
tic
al
m
o
d
e
l fo
r
robo
tic fish
in
term
s o
f
p
r
o
p
e
lling
and
an
gu
lar in
cli
n
atio
n
of fish
m
ovem
e
nt
i
s
p
r
esent
e
d i
n
t
h
e
wo
rk
s
of
Hy
ou
ng
Se
ok
Ki
m
and
hi
s
col
l
eag
u
e
s (
2
0
0
7
)
an
d
Pi
chet
S
u
e
b
sai
p
r
o
m
and
hi
s col
l
e
a
gue
s (
2
0
1
2
).
K
o
r
k
m
az et
al
. (20
1
1
)
has i
n
ve
stigated the re
sistant speed
c
ont
rol o
f
fis
h
r
o
b
o
ts.
Afolayan Matthew Olatunde
pres
en
ted d
e
sig
n
of a stab
le co
n
t
ro
ller i
n
robo
tic fish.
The pat
h
control of a t
h
ree
-
wheel
ro
b
o
t
has
been i
nvest
i
g
at
ed usi
ng
o
p
t
i
m
i
zed PI
D co
nt
r
o
l
l
e
r wi
t
h
PSO al
go
ri
t
h
m
by
Tu
rki
Y.
Ab
dal
l
a
an
d hi
s col
l
eag
ue (
2
01
2
)
. I
n
t
h
i
s
s
t
udy
, a P
I
D c
ont
rol
l
e
r
was
use
d
t
o
co
n
t
ro
l two
featu
r
es
o
f
th
e
robo
t in
clud
ing
speed
co
n
t
ro
l and
co
n
t
ro
lling
th
e turn
i
n
g ang
l
e o
f
th
e
wheels. PSO
alg
o
rith
m
was u
s
ed
to find
t
h
e op
ti
m
a
l co
efficien
ts o
f
th
e co
n
t
ro
ller.
Pie-Jun Lee and his colleag
ue
s (2
01
2) i
n
t
r
o
d
u
ces an ap
pl
i
c
at
i
on
of Fuzzy
logic in the design of an
in
tellig
en
t fish
ro
bo
t with
m
u
ltip
le act
u
a
tors. Th
is
ro
bo
t is cap
able o
f
swimmi
n
g
easily and actin
g
inde
pende
n
tly in the
face
of
a
n
y haza
rd in water.
M
a
no
j Kus
h
w
a
h
and hi
s
c
o
l
l
eague (
2
0
1
4
)
c
o
m
p
ared t
h
e
di
f
f
ere
n
t
m
e
t
hods of
a
d
j
u
st
i
n
g
t
h
e
param
e
t
e
rs of
PID
co
nt
r
o
l
l
e
r
usi
n
g s
o
ft
co
m
put
i
ng t
ech
n
i
ques
suc
h
as
Genet
i
c
s,
PS
O
an
d F
u
zzy
i
n
DC
m
o
to
r. Th
e
resu
lts ind
i
cate the priv
ileg
e
of
ev
o
l
u
tio
n
a
ry
al
gorithm
s
to classical m
e
thods suc
h
as Zie
g
ler or
Nichols. They
also indicated a pr
i
v
ilege of Fuzzy techniques to PI
D controller. Howe
ve
r, the Fuzzy
co
n
t
ro
ller is m
o
re co
m
p
licate
d
in
stru
ct
u
r
e
an
d co
stly as com
p
ared
to
th
e
PID con
t
ro
ller. If
PID co
n
t
ro
l
l
er is
ad
ju
sted
to enhan
ce its
p
e
rforman
ce, it wou
l
d
b
e
m
o
re
bene
ficial than t
h
e
Fuzzy c
ont
roll
er.
Prior
to
t
h
e p
r
esen
t
stud
y, n
o
p
r
ev
iou
s
research
h
a
s
b
een
con
d
u
c
ted, reg
a
rd
ing
co
n
t
ro
lling
rob
o
t
fish
usi
n
g
opt
i
m
i
z
ed P
I
D
co
nt
r
o
l
l
e
rs
based
o
n
e
vol
ut
i
ona
ry
al
go
rith
m
s
; th
u
s
t
h
is stud
y seems so
si
g
n
i
fican
t
in
th
is
p
o
i
n
t
th
at it app
lies evo
l
u
tionary algo
rith
m
o
n
th
e
ro
bo
t fish
.
The
prese
n
t
re
search
at
t
e
m
p
t
s
t
o
de
vel
op a
ro
b
o
t
fi
sh m
a
d
e
u
p
o
f
3 pa
rt
s
an
d 2 act
uat
o
r
l
i
nks.
Thi
s
fish
co
nsists of th
ree
p
a
rts: a h
ead, a flex
ib
le p
a
rt
and
a rear tail. Th
e co
n
t
ro
lling
p
a
rts of th
e syst
e
m
are
applied to the flexible part
of the robot. In the followi
ng se
ction, th
e suggested
m
e
thod
whic
h is the use of an
o
p
tim
ized
PID con
t
ro
ller
v
i
a th
e
b
acterial fo
rag
i
ng
alg
o
rith
m
is in
trod
u
c
ed
for
co
n
t
ro
lling
fish
rob
o
t
trajectory.
Di
ffe
re
nt
sect
ions
of t
h
i
s
pa
p
e
r are or
ga
ni
ze
d as fol
l
o
ws. S
ect
i
on 2 i
n
vest
i
g
at
es t
h
e fi
sh r
o
b
o
t
m
odel
.
In S
ect
i
on
3, t
h
e de
si
g
n
o
f
P
I
D c
o
nt
roller i
s
elaborated. Sec
t
i
on 4
desc
ri
bes t
h
e
bact
eri
a
l
fora
gi
n
g
al
g
o
ri
t
h
m
.
Ex
peri
m
e
nt
al
resul
t
s
are
pres
ent
e
d i
n
sect
i
o
n 5 i
n
o
r
der t
o
anal
y
ze t
h
e su
gge
st
ed m
e
t
hod. E
v
e
n
t
u
al
l
y
, sect
i
o
n
6 prese
n
t
s
t
h
e concl
u
si
ve
rem
a
rks
.
2
.
T
H
E PROPOSED
METHOD
2.
1. Fi
sh Ro
bo
t
S
y
ste
m
Mo
d
e
l
i
n
g
In
ge
ner
a
l
,
a
f
i
sh s
w
i
m
s i
n
wat
e
r
by
t
w
o
ki
n
d
s
of
b
o
d
y
an
d/
o
r
ca
udal
fi
n
(B
C
F
-st
y
l
e
)
or
m
e
di
an
and/
or
pai
r
e
d
f
i
n (M
P
F
-st
y
l
e
)
.
I
n
fact
, a fi
s
h
pr
o
v
i
d
es
its
p
r
o
p
u
l
siv
e
fo
rce
in
th
ese t
w
o ways. Similarly, in
th
e
desi
g
n
of
fi
sh
ro
b
o
t
s
,
we t
r
y
t
o
p
r
o
d
u
ce t
h
e
pr
o
pul
si
ve
f
o
rc
e by
t
h
e
d
o
rsal
fi
n a
n
d t
h
e
an
al
fi
n
or t
h
e
fl
exi
b
l
e
p
a
r
t
o
f
th
e
ro
bo
t bod
y wh
ich
h
a
s t
h
e ro
le
o
f
th
e br
aw
n m
u
scles o
f
f
i
sh
[
19].
2.
1.
1. Fi
sh Ro
bot
D
y
n
a
mi
cs and
Ki
nem
a
ti
cs
Kin
e
m
a
tics is
th
e scien
ce
o
f
m
o
t
i
o
n
reg
a
rd
l
e
ss of th
e cau
s
es o
f
t
h
at m
o
ti
o
n
. Th
is science, in
fact
,
deals wit
h
s
p
a
ce, vel
o
city, accelerati
on a
n
d
all highe
r-orde
r de
rivative
s
.
Dyna
m
i
cs is an e
x
tended bra
n
ch of
me
c
h
a
n
i
c
s
w
h
ic
h
l
o
o
k
s
in
t
o
th
e
forces a
n
d c
a
uses m
o
tion.
Fi
gu
re
1 an
d
2
i
ndi
cat
e t
h
e st
ruct
ure
of
a fi
s
h
r
o
bot
a
n
d i
t
s
4 sect
i
o
ns
[1]
.
As i
t
can
be
ob
serve
d
,
t
h
e
robo
t con
s
ists
o
f
3 lin
ks.
Three servo
-
m
o
to
rs with en
cod
e
rs are lo
cated at th
e fi
rst, secon
d
and th
ird
li
n
k
of
f
l
ex
ib
le bod
y p
a
r
t
of
th
e f
i
sh
r
obo
t, r
e
sp
ectiv
ely. I
n
th
e p
r
esen
t research, th
e first an
d
seco
nd
lin
k
s
are
actuators
while the thi
r
d is
stable. T
h
e
fourth piece is t
h
e tail whic
h is
use
d
as a
propeller
to produce a
gentle
m
ove
m
e
nt similar to that
of a
real fis
h
a
n
d is
attached
t
o
th
e
th
ird lin
k.
Here, th
ere are th
e th
ree m
a
in
p
a
rt
s:
1)
Head
(
r
o
b
o
t
ri
gi
d
pa
rt
)
2)
Flex
ib
le p
a
rts
3)
Th
e ch
ip h
a
ng
in
g fro
m
th
e flex
ib
le
p
a
rts
(fish
tail)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1372 –
1380
1
374
Figure 1.
Struc
t
ures of
a
fish
robot with 4 pieces
and
3 l
i
n
ks
[1]
Figure 2.
Struc
t
ures of
a
fish
robot with 4 pieces
and
3 l
i
n
ks
Fish
r
obo
t m
e
c
h
an
ical con
s
tr
ain
t
s ar
e show
n in
Figu
r
e
3 and
th
e
k
i
n
e
m
a
ti
c p
a
r
a
m
e
ter
s
of
f
i
sh
ro
bo
t
dynam
i
c
m
odel are indicated in Fi
g
u
re 4. I
n
these fig
u
res
,
T
1
and T
2
are
respectively the input t
o
rques for
ro
tating
th
e
1
st
and
2
nd
l
i
nks
.
I
i
,
l
i
, and
a
i
are
resp
ectiv
ely mo
m
e
n
t
s o
f
in
ertia o
f
th
e i
th
l
i
nk, i
th
lin
k
leng
t
h
and
the dista
n
ce
be
tween t
h
e
body
center and t
h
e
i
th
j
o
in
t [1
].
Fi
gu
re
3.
Fi
sh
R
o
b
o
t
M
echa
n
i
cal
C
onst
r
ai
nt
s
Fi
gu
re
4.
B
a
si
c M
o
d
e
l
o
f
t
h
e
Fi
sh R
o
b
o
t
M
O
TI
O
N
F
F
and
F
C
i
n
di
cat
ed i
n
fi
gu
re
5 re
pre
s
ent
t
h
e pr
o
pul
si
on
force a
n
d lift force
pr
odu
ced
as a r
e
su
lt of
t
h
e m
ovem
e
nt
of
t
h
e
3
rd
joi
n
t
cause
d by
a hy
dra
u
l
i
c
i
n
t
e
ract
i
on [
3
]
.
Fi
gu
re 5.
F
F
and F
C
of Tai
l
Fi
n [3]
Our assu
m
p
tio
n
is th
at th
e i
n
ertia force an
d
th
ru
st fo
rce app
l
y th
e tail fin
.
Th
erefo
r
e, F
F
is app
lied
on
th
e who
l
e tail fin
and
F
C
on
the tail f
i
n
.
Th
e dyn
amic eq
u
a
tion
f
o
r
each
r
obot lin
k
is stated
i
n
r
e
latio
n
(
1
)
[3
].
G
G
G
τ
τ
τ
(1)
Param
e
ters H
11
to
H
33
and G
1
to
G
3
as well
as T
1
to
T
3
i
n
thi
s
eq
uat
i
on ar
e
m
e
nt
i
oned i
n
[3]
,
[1
7]
-
[1
8]
.
In ca
se the
fish robot is consi
d
ere
d
as a
ri
gid body
, t
h
e
pr
op
ul
si
o
n
eq
uat
i
o
n
of
t
h
e
ro
b
o
t
ca
n
be st
at
ed
(2
) [3]
:
F
F
(2)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Op
timized
PID Con
t
ro
ller with
Ba
cteria
l Fora
g
i
ng
Algo
rit
h
m (S
eiyed Moh
a
mmad
Mirzaei
)
1
375
Wh
ere, M is m
a
ss of th
e
bo
d
y
,
x
is th
e
po
sitio
n of th
e cen
t
er of
b
o
d
y
’s m
a
ss, and
F
F
and
F
D
are the
thrust and
th
e
to
tal d
r
ag
fo
r
ces
, respecti
v
ely.
2.
2.
Fluid F
o
r
ce Model
If we c
onsi
d
er
a fi
sh ro
bot
i
n
a fl
ui
d wi
t
h
a con
s
t
a
nt
fl
o
w
of
U
m
, th
en
the in
ertial fo
rce an
d
th
e lift
force on
its tail
fin
can
b
e
ob
tain
ed
.
In
th
is case, th
e th
ru
st fo
rce (F
F
) an
d l
a
teral force
(F
C
) com
pone
nt
s c
a
n be
calcu
lated
[20
]
. Fig
u
r
e
6
in
dicates th
e in
ertial flu
i
d
fo
rce
wh
ile Figu
re 7 sh
ows th
e lift
force on
th
e tail fin
.
Here
,
U
is th
e relativ
e v
e
lo
city at
th
e cen
tre of th
e tail fin
and
α
is t
h
e attack angl
e of the
robot, i.e.,
swi
m
m
i
ng st
ar
t
angl
e
of t
h
e c
a
udal
fi
n i
n
t
h
e ro
b
o
t
.
Al
s
o
,
2C
i
s
t
h
e c
h
or
d l
e
n
g
t
h
ge
ner
a
t
e
d by
t
h
e
t
a
i
l
of t
h
e
fish
i
n
water, L is th
e sp
an
o
f
tail fin
,
and
ρ
is th
e
d
e
nsity o
f
water
[3
].
A
s
it can
be ob
serv
ed
in f
i
gu
r
e
6, F
V
is
force
p
r
o
portio
nal to
th
e acc
eleration acted i
n
the
opposit
e
d
i
rection
o
f
t
h
e fish
tail, an
d
F
is th
e lift fo
rce in
th
e
p
e
rp
end
i
cu
lar
d
i
rection
of ca
udal fin
whic
h is estimated
in
Eq
u
a
tion
(3
) [4
].
Fi
gu
re 6.
I
n
ert
i
al
Fl
ui
d Fo
rce [3]
Fi
gu
re 7.
Li
ft
For
ce
[
4
]
F
J
is stated i
n
Euation (3):
F
2
.
(
3
)
There
f
ore, F
F
an
d F
C
ca
n
be st
at
ed i
n
Eq
uat
i
o
n
(4
) a
n
d
(
5
)
[
4
]
.
.s
i
n
.s
i
n
(4)
.
cos
.
cos
(5)
2.
2.
1.
M
o
ti
on
E
qua
ti
on
o
f
F
i
sh R
o
b
o
t
If we as
sum
e
that the fish
robot m
oves in
x
-d
irection
,
th
en
th
e relativ
e v
e
l
o
city at th
e
m
a
ss cen
tre
o
f
th
e tail fin
i
n
y
-d
irection
is esti
m
a
ted
th
rou
g
h
Equ
a
tio
n (6
)
[4
].
cos
.
cos
.
(6)
Whe
r
e
U
m
is co
n
s
tan
t
flow,
u is relativ
e v
e
lo
city at th
e cen
ter of th
e
fish tail in
Y
m
direction. T
h
ere
f
ore, the
relativ
e v
e
l
o
city (
U
) is stated
as (7):
(7)
Fin
a
lly, If Eq
uatio
n
(3) an
d (7
) are i
n
serted
in
(1
), m
o
tio
n
eq
u
a
tion
of th
e fish rob
o
t
is
ob
tain
ed as
fo
ll
o
w
s:
N
N
N
(8)
In th
is
relatio
n
,
M
11
to
M
33
and
N
1
to N
3
a
r
e
m
e
nt
i
oned
i
n
r
e
fere
nce[
3]
.
The
SIM
U
LI
N
K
bl
oc
k
di
ag
ra
m
of t
h
e fi
s
h
r
o
b
o
t
m
o
t
i
on m
odel
i
s
p
r
ese
n
t
e
d i
n
Fi
g
u
r
e
8.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1372 –
1380
1
376
Fi
gu
re
8.
SIM
U
LI
N
K
B
l
oc
k
Di
ag
ram
of t
h
e
Fi
sh R
o
bot
M
o
t
i
o
n
M
o
del
U
s
i
n
g
PI
D C
ont
r
o
l
l
e
r
2.3. PID
Controller
PID
co
nt
r
o
l
l
e
r
i
s
am
ong t
h
e
m
o
st
preval
ent
i
n
st
ances
o
f
fe
edbac
k
c
o
nt
rol
l
ers use
d
i
n
a v
a
st
m
a
jori
t
y
of c
ont
r
o
l
p
r
o
cesses suc
h
as
cont
r
o
l
l
i
ng
D
C
m
o
t
o
r vel
o
c
i
t
y
, pressu
re c
ont
rol
,
t
e
m
p
erat
ure co
nt
r
o
l
,
et
c. The
goal
o
f
usi
ng
PID al
g
o
r
i
t
h
m
i
n
t
h
e cl
osed-
l
oo
p co
nt
r
o
l
i
s
t
h
e preci
se an
d fast
co
nt
rol
l
i
ng
of sy
st
em
out
put
un
de
r va
ri
o
u
s
ci
rcum
st
ances and
wi
t
h
out
a
det
a
i
l
e
d k
n
o
w
l
e
dge
of
sy
st
em
behavi
or i
n
resp
o
n
se t
o
t
h
e i
n
p
u
t
.
PID c
o
nt
rol
l
e
r
i
s
co
m
p
ri
sed
of
3 di
st
i
n
ct
p
a
rt
s:
pr
op
ort
i
o
n
al
, i
n
t
e
g
r
al
and
deri
vat
i
v
e.
Each w
o
ul
d t
a
ke t
h
e
sig
n
a
l error as
th
e in
pu
t and
pro
cess it. Fin
a
lly, th
eir o
u
t
put
s are summ
ed
up. The
outp
ut
of t
h
i
s
sy
st
em
whi
c
h
is th
e sam
e
as PID ou
tpu
t
is
sen
t
to
t
h
e system
fo
r error co
rrectio
n.
PID con
t
ro
ller
o
u
t
p
u
t
is estimated
in
Equ
a
tion
(9):
u(t
)
=
(9)
Varia
b
le
e
sta
n
d
s
for th
e track
i
ng
error which
is th
e d
i
ffe
rence
of the target val
u
e (r) and the
real
v
a
lu
e
o
f
ou
tpu
t
(y).
Th
e error
sig
n
a
l en
ters t
h
e con
t
ro
ller and
th
e d
e
riv
a
tive and
in
teg
r
al
v
a
lu
es are esti
mated
.
The
n
t
h
e cont
r
o
l
si
gnal
u
is estim
a
ted using a coefficient of error signal (K
p
), a coe
ffi
cient of error integral
(K
i
) a
n
d a c
o
efficient of e
r
ror
deri
vative
(K
d
).
B
a
sed o
n
t
h
e
abo
v
e
-
m
e
nt
i
oned i
ssue
s
, i
t
can be
r
ealized that a control
syste
m
requires tuning.
Gen
e
rally sp
eak
i
ng
, wh
at is esti
m
a
ted
in
th
e d
e
sign
ph
ase
sh
ou
l
d
no
t lead
u
s
to
ex
p
e
ct th
e sam
e
fin
d
i
n
g
s
i
n
practice. Tuning is a significant issue.
PID co
efficien
t n
eeds to
b
e
altered
so
m
a
n
y
ti
mes
th
at th
e resu
lts o
f
an
opt
i
m
i
zed resp
ons
e are o
b
t
a
i
n
ed
. The
r
e exi
s
t
s
a vari
et
y
of m
e
t
hods
fo
r
t
uni
n
g
.
One s
u
ch m
e
t
h
o
d
i
s
usi
n
g
evol
ut
i
ona
ry
al
go
ri
t
h
m
s
.
2
.
4
.
PID
Co
ntro
ller
fo
r Contro
lling
Fish Robot Motion
In t
h
i
s
sect
i
o
n,
a PI
D c
o
nt
rol
l
er i
s
desi
g
n
ed
fo
r t
h
e
fi
s
h
m
o
t
i
on.
In
t
h
i
s
de
si
gn
, 4
pa
rt
s a
r
e co
nt
ri
ve
d
of
f
o
r t
h
e
fi
sh
ro
b
o
t
.
Fi
gu
re
9
sh
ow
s t
h
e
st
r
u
ct
ure,
di
rect
i
o
n
an
d c
ont
rol
of
t
h
e fi
s
h
ro
b
o
t
m
o
ti
on
usi
n
g a
PI
D
cont
rol
l
e
r.
Fi
g
u
re
1
0
s
h
ows
S
I
M
U
L
I
N
K
bl
o
c
k
di
ag
ram
of f
i
sh r
o
bot
usi
n
g
PI
D c
ont
rol
l
e
r
s
.
Fi
gu
re
9.
St
r
u
c
t
ure,
Di
rect
i
o
n
an
d Con
t
ro
l of
th
e Fish
Ro
bo
t
Using
PID C
o
n
t
ro
ller
Fi
gu
re
1
0
.
SIM
U
LI
N
K
B
l
oc
k
Di
ag
ram
of Fi
s
h
R
o
b
o
t
Using
PID Con
t
ro
llers
In
o
r
d
e
r to
tune PID con
t
ro
ller p
a
ram
e
ters i
n
th
is stud
y, t
w
o
al
g
o
rith
m
s
were sug
g
e
sted
wh
ich
were
b
o
t
h
b
a
sed
o
n
swarm
in
tellig
en
ce: b
a
cterial forag
i
n
g
algo
ri
th
m
,
an
d
PSO
alg
o
rith
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Op
timized
PID Con
t
ro
ller with
Ba
cteria
l Fora
g
i
ng
Algo
rit
h
m (S
eiyed Moh
a
mmad
Mirzaei
)
1
377
2.
5.
B
a
cteri
al
For
agi
n
g
Al
g
o
r
i
t
hm
Bacterial forag
i
ng
algorithm
is a
g
l
ob
al
op
ti
m
i
zatio
n
alg
o
rith
m
su
gg
ested in
[1
3
]
. Th
e id
ea of
b
acterial fo
rag
i
n
g
algo
rith
m
is b
a
sed
o
n
th
e
fact th
at in
n
a
ture, liv
e creatures with
po
o
r
er
fo
ragi
ng
strate
gy
are
m
o
re pro
n
e t
o
ext
i
n
ct
i
on t
h
a
n
t
h
o
s
e wi
t
h
a
successf
ul
f
o
r
a
gi
n
g
st
rat
e
gy
.
Aft
e
r a m
u
l
t
itude
of
gene
rat
i
ons
,
creature
s
with poorer fora
ging
strate
gies are
either e
x
tinct
or t
r
ans
cen
de
d
t
o
m
o
re de
vel
ope
d s
p
eci
es.
E-coi
l
bact
eri
u
m
whi
c
h d
w
el
l
s
i
n
hum
an i
n
t
e
st
i
n
es has a
3-st
age f
o
ra
gi
n
g
m
e
t
hod:
chem
ot
axi
s
(t
um
bl
ing a
n
d
swi
m
m
i
ng),
re
pr
o
duct
i
o
n,
an
d el
i
m
i
n
at
i
on-
d
i
spersal
.
Chem
ot
a
x
is
: In
th
is stag
e, the b
acteri
u
m
b
e
g
i
n
s
to
t
u
m
b
le
and
swim
d
e
pen
d
i
n
g
on
t
h
e
ro
tation
of
bact
eri
a
’s t
a
i
l
.
If t
h
e am
ount
of f
o
o
d
i
s
m
o
re al
o
n
g
t
h
e new
di
rect
i
o
n,
bact
eri
a
cont
i
nue
s t
o
m
ove i
n
t
h
at
di
rect
i
o
n (
s
wi
m
m
i
ng).
No
w s
u
p
p
o
se
that we wa
nt t
o
fi
nd t
h
e m
i
n
i
m
u
m
of J(
Ɵ
) w
h
er
e
Ɵ
ϵ
R
r
.
Ɵ
is th
e
p
o
s
i
tio
n
of the
bacterium
.
J(
Ɵ
) i
s
i
n
di
cat
i
v
e
of t
h
e am
ount
of
fo
o
d
i
n
t
h
e l
o
cat
i
o
n
o
f
Ɵ
. J
(
Ɵ
)>0
,
j(
Ɵ
)=
0,J
(
Ɵ
)<0 m
eans t
h
at the
bact
eri
u
m
i
s
r
e
spect
i
v
el
y
wi
t
h
s
u
f
f
i
c
i
e
nt
,
neut
ral
an
d
i
n
suf
f
i
c
i
e
nt
am
ount
o
f
fo
o
d
i
n
l
o
cat
i
o
n
Ɵ
. Fo
r the
tum
b
le to take place, a
vector
in the ra
ndom
direction ca
lled Ø(i)
whose e
l
e
m
ents lie in [-1, 1] is create
d
. T
h
is
vect
o
r
i
s
use
d
fo
r defi
ni
n
g
t
h
e
ne
w di
rect
i
o
n of
t
u
m
b
le for th
e b
a
cteriu
m on
ce it is
started
.
Th
e new po
siti
on
of
t
h
e
bact
eri
u
m
i
s
defi
ne
d
as
rel
a
t
i
o
n
(
1
1
)
:
Ɵ
(
j
+1,k,l)
=
Ɵ
(
j
,k,l)
+ C(
i)
Ø(i)
(11)
Here
,
Ɵ
i
(j,k,l) indicates the i
th
bact
eri
u
m
posi
t
i
on i
n
t
h
e j
th
chem
otactic, th
e k
th
re
pr
od
uct
i
on a
nd t
h
e
l
th
el
i
m
in
atio
n
an
d
d
i
sp
ersal. C(i)
is
th
e
size
of t
h
e bacteri
u
m tu
m
b
le in the dir
ection
of
Ø(i). If th
e size o
f
J(I,
j
,
k
,
l) in
Ɵ
i
(j+1
,k
,l) is sm
al
l
e
r th
an
its size in
Ɵ
i
(j
,k
,l), then
a
furth
e
r tum
b
le is
mad
e
in
th
e size of C
(
i) and
in
th
e
d
i
rection o
f
Ø(i).
Th
en
th
e b
acteriu
m
wou
l
d b
e
g
i
n to
swim
in
th
e d
i
rectio
n of
Ø(i). As l
o
ng
as th
e size
of J(
Ɵ
) i
s
bei
n
g
decrease
d
,
a
n
d
u
p
t
o
t
h
e m
a
xi
m
u
m
num
b
e
r o
f
s
w
i
m
m
i
ng al
l
o
wed
(
N
s
) t
h
i
s
swi
m
m
i
ng
w
oul
d
co
n
tinu
e
. Th
is in
d
i
cates t
h
at
th
e
b
acteri
u
m con
tinu
e
s its
d
i
rection
u
n
til it find
s a b
e
t
t
er env
i
ron
m
e
n
t fo
r
fo
ragi
ng
.
Repr
oduc
tion
: The least hea
lthy bacteria ev
ent
u
ally die while each
of th
e healthie
r ba
cteria (Thos
e
yield
i
n
g
lower v
a
lu
e of
th
e obj
ectiv
e fun
c
tion
)
asexu
a
lly
split in
to
two b
a
cteria, wh
ich
are th
en p
l
aced
i
n
th
e
sam
e
lo
catio
n
.
Th
is k
eep
s the
swarm
size constant.
Elimina
t
io
n and
dispersi
o
n
:
The l
i
f
e o
f
a
po
p
u
l
a
t
i
on
of
bact
eri
a
i
s
g
r
a
dual
l
y
al
t
e
red t
h
r
o
ug
h f
o
od
co
nsu
m
p
tio
n
or
su
dd
en
ly
thro
ugh
o
t
h
e
r facto
r
s.
Ev
en
ts
can
cau
s
e th
e b
acteria to
be k
illed
or
d
i
sp
ersed
.
Although, initi
ally this
m
i
ght disrupt
t
h
e foraging
sta
g
e,
it can have
a pos
itive effect as
well. That is
because
bact
eri
a
di
spe
r
si
on
co
ul
d
kee
p
t
h
em
close to s
p
ots
whe
r
e
an a
b
unda
nce
o
f
fo
od is av
ailed
.
Th
e eli
m
i
n
ation
and dis
p
ersi
on
stage would
prevent t
h
e
bacteria to
be tra
p
pe
d in the l
o
cal
optim
i
zed poi
nt. In each elim
ination
an
d d
i
sp
ersion stag
e, th
e
pro
b
a
b
ility o
f
eli
m
in
atio
n
and d
i
sp
ersi
o
n
for an
y resi
d
i
ng b
acteri
u
m
in
th
e
p
opu
latio
n
is P
ed
. In
or
der t
o
kee
p
t
h
e n
u
m
ber of bact
e
r
i
a
c
onsta
nt, if a bacterium
is eliminated, a new
bacterium
is ra
ndom
ly replaced in the
searc
h
ing z
one
.
3.
RESULTS
A
N
D
DI
SC
US
S
I
ON
Fi
gu
re 1
1
i
ndi
cat
es t
h
e st
ruct
ure o
f
t
h
e st
u
d
i
ed fi
sh r
o
b
o
t
com
p
ri
si
ng o
f
3 l
i
nks. Li
nk
1
and l
i
nk
2
have
m
o
t
o
r a
n
d e
n
co
de
r re
sp
ect
i
v
el
y
and
m
o
t
o
rs
gene
rat
e
T
1
and T
2
.
H
o
weve
r,
Li
n
k
3
has
n
’t
act
uat
o
r a
n
d
enco
de
r. T
h
e
p
a
ram
e
t
e
rs of si
m
u
l
a
t
i
on are
s
h
o
w
n
by
Ta
bl
e 1.
Fi
gu
re 1
1
. St
ru
ct
ure of
Fi
s
h
R
o
b
o
t
[3]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1372 –
1380
1
378
Tabl
e 1. Param
e
t
e
rs
f
o
r
Si
m
u
lat
i
on [3]
1.23
1
0
1.6
0.35
0.175
M
o
m
e
nt of I
n
er
tia
Mass
L
e
ngth
Center
of Gr
avity
L
i
nk 1
3.7
10
1
0.115
0.0575
M
o
m
e
nt of I
n
er
tia
Mass
L
e
ngth
Center
of Gr
avity
L
i
nk 2
6.75
1
0
0.33
0.062
0.031
M
o
m
e
nt of I
n
er
tia
Mass
L
e
ngth
Center
of Gr
avity
L
i
nk 3
998
0
.
5
Wate
r Densit
y
drag Coef
f
i
cient
Fluid
Force
In the e
xpe
riments, PID c
ont
roller coefficients
are estim
a
ted running
the PSO algorithm
.
The
obtaine
d c
o
e
ffi
cients are:
= 5
.
45
10
0
.
1210
=
1
.
51
50
Also
PID con
t
ro
ller co
efficients u
s
ing
t
h
e Bacterial Fo
rag
i
ng
algorith
m
is as fo
llo
ws:
= 6.33
20
0
.
1210
= 1.82
87
MATLAB/SIMULINK is used
to
co
m
p
are th
e o
p
tim
iz
e
d
PID tho
ugh
Bacterial Fo
rag
i
ng
algo
rith
m
an
d Fu
zzy co
ntro
ller
[1
] and
yet with
th
e op
ti
m
i
zed
PI
D t
h
r
o
ug
h
PS
O al
go
ri
t
h
m
[1
6]
.
A c
ont
rol
sy
st
em
i
s
d
e
sirab
l
e wh
en u
pon
th
e en
tran
ce
o
f
an
inpu
t, it m
a
n
a
g
e
s to
track
it
with th
e fewest erro
r and
in
m
a
x
i
m
u
m
length
of ti
m
e
. The less overshoot and settling ti
m
e
for
the output, and the soone
r th
e final state
is reached,
the better perform
ance the cont
rol system
will have. In
this study, two inputs ha
ve been applied on the
syste
m
. O
n
e is
th
e step stand
a
r
d
inp
u
t
and
ano
t
h
e
r
is th
e sine stand
a
rd
input.
Fi
gu
re
1
2
s
h
o
w
s t
h
e
desi
re
d
m
easured
p
o
s
i
t
i
on o
f
fi
sh
r
o
bot
’s
1
st
link
(
Ɵ
1
)
u
s
i
n
g 3 con
t
ro
llers: an
o
p
tim
ized
PI
D th
r
oug
h
t
h
e Bacter
ial Fo
r
a
g
i
n
g
al
g
o
r
ith
m
,
Fu
zzy con
t
ro
ller
and
th
e
o
p
t
i
m
ized
PI
D
through
PSO algo
rith
m
with
app
l
yin
g
step
inpu
t.
Fi
gu
re 1
3
s
h
o
w
s
t
h
e desi
re
d m
easured
p
o
si
t
i
on o
f
fi
s
h
ro
b
o
t
’
s 2
nd
link
(
Ɵ
2
)
w
ith app
l
ying
step
inpu
t.
Fig
u
r
e
12
. Resp
on
se to
t
h
e St
ep
I
npu
t in
R
o
b
o
t
’
s
1
s
t
Link (
Ɵ
1
)
Fig
u
r
e
13
. Resp
on
se to
t
h
e St
ep
I
npu
t in
R
o
b
o
t
’
s
2
n
d
Link (
Ɵ
2
)
The
n
t
h
e si
ne i
n
p
u
t
ent
e
red t
h
e sy
st
em
. The reaso
n
was t
h
e
qua
si
-si
n
e m
ovem
e
nt
of fi
s
h
.
Fi
gu
re
1
4
sh
ows of t
h
e
desired m
easu
r
ed
p
o
sitio
n
of fish
robo
t’s
1
st
lin
k (
Ɵ
1
) us
ing
th
e
th
r
e
e c
o
n
t
rolle
r
s
wh
en
ap
plyin
g
sin
e
in
pu
t.
A
l
so
, Figur
e 15
sh
ow
s th
e
d
e
sired
m
easu
r
ed
po
sitio
n
o
f
f
i
sh
r
obo
t’
s 2
nd
link
(
Ɵ
2
) using
the th
ree
cont
rollers
with applying si
ne
input.
While the sine input was used,
still
the optim
ized PI
D controller through
bact
eri
a
l
fo
ra
g
i
ng al
g
o
ri
t
h
m
sho
w
e
d
t
o
ha
v
e
a bet
t
e
r per
f
o
rm
ance t
h
an i
t
s t
w
o co
unt
e
r
part
s
bot
h i
n
t
r
acki
n
g
the input t
r
ajec
tory a
n
d
m
i
nim
i
zi
ng si
g
n
al
e
r
r
o
r
.
0
2
4
6
8
10
12
14
16
18
20
-0
.
5
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
In
pu
t
Re
fr
e
n
c
e
f
u
z
zy co
n
t
r
o
l
l
e
r
O
p
t
i
m
i
z
ed P
I
D c
ont
r
o
l
l
e
r
w
i
t
h
P
S
O
O
p
t
i
m
i
z
ed P
I
D c
ont
r
o
l
l
e
r
w
i
t
h
B
F
0
2
4
6
8
10
12
14
16
18
20
-0
.
5
0
0.
5
1
1.
5
2
2.
5
I
npu
t
R
e
f
r
e
n
c
e
f
u
zz
y c
o
n
t
r
o
l
l
e
r
O
p
t
i
m
i
z
e
d P
I
D
c
ont
rol
l
e
r
w
i
t
h
P
S
O
O
p
t
i
m
i
z
e
d P
I
D
c
ont
rol
l
e
r
w
i
t
h
B
F
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Op
timized
PID Con
t
ro
ller with
Ba
cteria
l Fora
g
i
ng
Algo
rit
h
m (S
eiyed Moh
a
mmad
Mirzaei
)
1
379
Fig
u
r
e
14
. Resp
on
se to
t
h
e si
n
e
inpu
t in
ro
bo
t’
s
1
s
t
l
i
nk (
Ɵ
1
),
Fig
u
r
e
15
. Resp
on
se to
t
h
e Si
n
e
I
npu
t in
R
o
b
o
t
’
s
2
n
d
Link (
Ɵ
2
)
Table 2 i
ndica
tes the analytic feature
s
of
th
e respo
n
s
e to step
in
pu
t in
ro
bo
t’s
1
st
l
i
nk
(
Ɵ
1
) a
n
d 2
nd
l
i
nk (
Ɵ
2
). Th
is tab
l
e p
r
ov
es th
at th
e op
tim
i
zed
PID con
t
ro
ller th
rou
g
h
t
h
e Bacterial Fo
rag
i
ng
algo
rit
h
m
h
a
s
had a
bet
t
e
r p
e
rf
orm
a
nce i
n
t
h
e 1
st
l
i
nk (
Ɵ
1
) co
m
p
ared to
th
e Fu
zzy
co
n
t
ro
ller and
th
e op
tim
iz
ed
PID
co
n
t
ro
ller t
h
rou
g
h
PSO al
g
o
rith
m
.
However, a similar resp
on
se is
o
b
serv
ed
on
th
e
2
nd
lin
k
wh
en
eith
er t
h
e
BF algo
rith
m
o
r
PSO al
go
rithm
is u
s
ed.
Tab
l
e
2
.
An
alysis o
f
Resp
on
se to
Step
I
npu
t
Over
shoot
Rise Ti
m
e
(sec
)
Settling Ti
m
e
(sec
)
Lin
k
Lin
k
30%
2
4.
5
Fuzzy
L
i
nk 1
10%
1.
5
2.
5
PID PS
O
5%
1.
5
2
PID BF
E
r
r
a
tic Undulations
1
4.
5
Fuzzy
L
i
nk 2
0%
1.
75
2
PID PS
O
0%
1.
75
2
PID BF
Table 3 is indi
cative of the
mean
-squa
r
ed error
si
gnals between
th
e i
n
pu
t and
ou
tpu
t
for th
e sine
in
pu
t in
th
e 1
st
j
o
in
t (
Ɵ
1
) a
n
d
2
nd
j
o
in
t (
Ɵ
2
). Th
is tab
l
e p
r
ov
es th
at th
e PID con
t
ro
ller op
ti
m
i
zed
th
roug
h
th
e
bact
eri
a
l
f
o
ra
g
i
ng al
g
o
r
i
t
h
m
has
had a
bet
t
er f
unct
i
oni
ng com
p
ared to
the Fuzzy a
nd the PID c
ont
roller
o
p
tim
ized
th
rou
g
h
PSO algorith
m
.
Tabl
e
3.
A
n
al
y
s
i
s
of
R
e
s
p
o
n
s
e
t
o
Si
ne
In
p
u
t
M
ean Squar
e
Signal E
r
r
o
r
Lin
k
Lin
k
0.
1101
Fuzzy controller
L
i
nk 1
0.
0198
PID PS
O
0.
0140
PID BF
0.
0017
Fuzzy controller
L
i
nk 2
0.
0001
676
PID PS
O
0.
0001
073
PID BF
Am
ong t
h
e rea
s
on
s o
f
why
t
h
e fi
n
d
i
n
gs
of
B
F
al
go
ri
t
h
m
are cl
ose
r
t
o
r
eal
m
i
nim
a
com
p
ared wi
t
h
ot
he
r ap
pr
oach
es such as PS
O, i
t
can be
m
e
nt
i
one
d t
h
at
t
h
e searc
h
i
n
g m
e
t
hod i
n
B
F
al
go
ri
t
h
m
fol
l
o
ws l
o
ca
l
search. Bacteri
a
searc
h
in a
parallel and i
n
depende
n
t
way, an
d th
ere is no
in
terch
a
n
g
e
a
m
o
n
g
th
e
b
a
cteria.
Ho
we
ver
,
i
n
P
S
O al
g
o
r
i
t
h
m
,
al
l
part
i
c
l
e
s
m
ove
o
n
l
y
t
o
wa
r
d
s
one
best
par
t
i
c
l
e
(gbest
)
,
a
nd a
r
e
not
a
ffe
ct
ed b
y
any
of t
h
e ot
he
r part
i
c
l
e
s. Eve
n
i
f
a part
i
c
l
e
st
ands as t
h
e se
con
d
best
am
ong al
l
,
i
t
poses
no ef
fect
o
n
de
ci
si
on
mak
i
n
g
o
f
o
t
her p
a
rticles. Nev
e
rt
h
e
less, con
d
ition
s
are d
i
fferen
t
in BF.
A
t
th
e
reproductio
n
lev
e
l,
h
a
l
f
of th
e
bact
eri
a
a
r
e
o
m
i
t
t
e
d. T
h
e
ot
h
e
r
hal
f
pl
ay
s an
eq
ual
l
y
si
gni
fi
cant
r
o
l
e
i
n
p
r
o
duci
n
g
p
r
ospec
t
i
v
e res
p
o
n
ses
.
Also
, in BF al
g
o
rith
m
th
e num
b
e
r o
f
iteratio
n of ev
e
r
y
pa
rt
i
s
est
i
m
at
ed i
nde
pen
d
e
n
t
l
y
fr
om
ot
hers
.
Ho
we
ver
,
i
n
P
S
O al
g
o
r
i
t
h
m
,
t
h
e n
u
m
b
er o
f
i
t
e
rat
i
on f
o
r t
h
e ent
i
r
e al
g
o
ri
t
h
m
shoul
d be
defi
ned al
l
at
once
,
wh
ich
wou
l
d
resu
lt in
flex
ib
i
lity red
u
c
tio
n.
Bacterial Fo
rag
i
ng
algo
rith
m
,
th
oug
h
h
a
v
i
ng
sim
p
ler n
a
ture, is
found to
be m
o
re efficient a
n
d precise in
res
p
ons
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1372 –
1380
1
380
4.
CO
NCL
USI
O
N
As m
e
nt
i
oned
i
n
t
h
e p
r
evi
o
us sect
i
o
ns, t
h
e preci
si
o
n
of
a fi
sh r
o
bot
depe
n
d
s o
n
se
veral
fact
or
s
in
clu
d
i
n
g
t
h
e
p
r
ecision
o
f
m
o
t
i
o
n
sensors, lin
k
m
o
b
ilit
y, elasticit
y o
f
fish
rob
o
t
dri
v
ing
system
, a
n
d
t
h
e
p
r
ecision of con
t
ro
llers.
Th
e
m
o
st sig
n
i
fican
t
o
f
all is
the
precision
and efficiency
o
f
c
ont
rol
l
e
rs
. T
h
e
r
ef
ore
,
we ha
ve
use
d
PID c
o
nt
r
o
l
l
e
r,
t
h
at
i
s
one
of
t
h
e m
o
st
preva
l
ent
and
p
o
we
r
f
ul
c
ont
r
o
l
l
e
rs
use
d
i
n
i
n
d
u
st
r
y
, t
o
cont
rol
t
h
e
di
re
ct
i
on
of
t
h
e
fi
s
h
ro
b
o
t
.
T
w
o e
vol
ut
i
ona
ry
al
g
o
ri
t
h
m
s
were
s
u
g
g
est
e
d i
n
t
h
i
s
resea
r
ch
f
o
r t
uni
ng
PID
p
a
ram
e
ters fo
r con
t
ro
lling
th
e fish
robot traj
ecto
r
y.
These tw
o
w
e
re th
e b
acterial fo
rag
i
ng
alg
o
rithm an
d
PSO
algo
rithm
.
Th
en
th
e su
gg
ested
PID
co
n
t
ro
ller co
mp
ared
to th
e
po
werfu
l
Fu
zzy
controller. For thi
s
pu
r
pose
,
a fi
s
h
ro
bot
was de
s
i
gne
d wi
t
h
3
h
i
nge
d-l
i
n
ks a
n
d t
h
e
n
dy
nam
i
c
m
odel
was
d
e
t
e
rm
i
n
ed, t
h
e
m
o
ti
on
equation
of t
h
e robot is im
ple
m
ente
d by SIMULINK. Fi
nally, each of th
e pre
v
iously-mentioned controllers
was a
p
pl
i
e
d i
n
t
h
i
s
m
odel
.
T
h
r
o
ug
h a
p
pl
y
i
ng
st
ep
an
d si
ne i
n
p
u
t
s
whi
c
h a
r
e t
h
e
st
an
d
a
rd
i
n
put
s i
n
c
ont
rol
syste
m
s, th
e ou
tpu
t
d
i
r
ection w
a
s
o
b
s
erv
e
d. Test r
e
su
lts ind
i
cated
th
at the o
p
tim
ized
PI
D
con
t
ro
ller
thr
ough
the Bacterial F
o
ra
ging algorithm
could
track step
an
d
sin
e
in
pu
ts with
th
e
fewest errors an
d
in
th
e m
a
x
i
m
u
m
tim
e
as com
p
ared to Fuzzy controllers a
n
d op
ti
m
i
zed
PID co
n
t
ro
llers th
rou
g
h
PSO.
Op
tim
ized
PID throug
h th
e Bacterial Forag
i
ng
al
g
o
r
i
t
h
m
can be a
us
eful
m
e
t
hod
o
f
co
nt
r
o
l
l
i
ng a
fish
ro
bo
t m
o
t
i
o
n
.
It can
as
well b
e
u
s
ed
i
n
con
t
ro
lling
fish
robo
ts with
m
o
re j
o
in
ts o
r
tho
s
e m
o
v
i
n
g
in
pat
h
way
s
whi
c
h
have
o
b
st
acl
es. S
u
c
h
i
n
vest
i
g
at
i
ons
can
be
ap
peal
i
n
g
t
o
pi
cs o
f
fu
rt
he
r re
search
.
REFERE
NC
ES
[1]
Rahmat-Allah H
ooshmand, Ali
Akbar Na
siri,
an
d Mohammad Ataei, “Trajector
y Angle Control
of Fish-like Rob
o
t
Motion b
y
Usin
g Fuzz
y Con
t
roll
er”,
Internationa
l Journal of Rob
o
tics and
Autom
a
tion
, Vol. 27, No. 2
,
2012
.
[2]
Pei-Jun Lee, Meng-Shiue L
e
e, and
Ruei-Ch
a
n Wang, “A Fuzzy
Contro
l B
a
sed Roboti
c
Fish with Multip
le
Actuators”,
In
ter
national Journal
of Fuzzy
Systems
, Vol. 14, No. 1
,
2012
.
[3]
H
y
oung Seok K
i
m, B
y
ungR
y
o
n
g
Lee, and
RakJ
in Kim, “A
Stud
y
on
the Motio
n Mechanism of
Articulated
Fish
Robot”,
International Conference on
M
echa
t
ronics and Au
tomation
, Harbin
, Ch
in
a, 2007
.
[4]
Pichet Suebs
a
ipr
o
m, Chun-Liang
L, “F
ish-Tail
Modeling for
Fish Robot”,
Inter
national S
y
mposium on Computer
,
Consumer and Control
, 2012
.
[5]
Y. John and
R.
Langar
i
, “Fuzzy
Logic Intellig
en
ce”,
Con
t
rol and
Information
, NJ: Prentice Hall, 1
998.
[6]
K. Streitlen, G.
S. Triantaf
y
llo
u, M. S. Trian
t
af
y
llou
,
“Efficient Foil Propulsion Through Vortex Control”, AIAA
Journal,
Vol. 34
, pp
. 2315-2319
, 1996
[7]
J. Yu, S. Wang
and M. Tan,
“A Simplified Pro
pulsive Model
of Biomim
etic
Robot Fish and
Its Realization
”
,
Robotica
, Vol.
23,
pp. 101-107
, 2005
.
[8]
H. Hu,
“Biologically
Inspired
Design
of
Autonomous
Rob
o
tic Fish at Essex”,
Pr
oceed
in
gs
of
the IEEE
SMC UK-RI
Ch
apter Conference on
Advan
ces
in
Cyber
n
e
tic
Sys
t
ems
, pp. 1-8, 20
06.
[9]
J. Liu
and H. Hu, “A 3D Simulato
r fo
r Autono
mous Robotic
Fish”,
Internatio
nal Journal of
Automation
and
Computing 1
,
Vol. 1
,
No. 1, pp
. 42-50
, 2004
.
[10]
X. Ye, Y. Su, S. Guo, L. Wang,
“Desi
gn and Realization of
a R
e
mote Control
Centimeter-Scale
Robotic Fish
”,
Pr
oceed
ings
of
th
e I
EEE
/ASM
E In
ter
nationa
l
Conf
er
enc
e
o
n
Ad
vanc
ed
In
tell
igent
Me
chat
r
onics
, pp. 25-3
0
,
2008.
[11]
D. Korkmaz, G
.
OzmenKoca,
and
Z. H. Akp
o
lat, “Robust Forward Sp
eed
Control of
a R
obotic Fish”,
6th
International Ad
vanced
Techno
lo
gies Symposium
(
I
ATS’11
)
, Ela
z
ı
ğ
, Turk
ey
, 2011.
[12]
Afolay
an Matth
ew Olatunde, “Stability
of th
e C
ontrol Scheme of a Design
of a Robotic Fish”,
Scholar lin
k
Research Institute Journals
, 2013
.
[13]
K.
Pa
ssi
no,
“Biomi
mi
c
r
y
of ba
cter
ial foraging
for distribut
ed o
p
tim
izat
ion
and
control
”
,
IEEE Control Systems
Magazine,
Vol.
22, No. 3, pp. 52
–67, 2002
.
[14]
K
e
nned
y
, J
., Eb
erhart
, R., "
Pro
ceed
ings of IEE
E
International
Conferenc
e
on Neural Network
s
". pp.
1942–19
48,
1995.
[15]
Turki Y. Abdalla, Abdulkareem. A.
A., ”PSO-based Optimum Design of
PID Controller
for Mobile R
obot
Traj
ector
y Tr
ack
ing”.
Internation
a
l Journal of
Co
mputer Applications
. Vol. 47, No. 23
, 2012
.
[16]
Manoj Kushwah
and Prof. As
his Patra,
”Tun
ing PID Controller fo
r Speed C
ontrol of
DC
Motor Using Soft
Computing Techniques-A Review”,
Advan
ce
in
Electronic and
Electric Engin
e
ering
, Vol. 4
,
No
. 2, pp. 141-148
,
2014.
[17]
J. Yu, L.
Liu, an
d L. Wang, “D
y
n
amics and
control of tu
rning m
a
neuve
r
for b
i
o
m
imetic robotic
fish”,
In
t. Conf.
on
Intelligen
t Robo
ts and Systems
, B
e
ijing
,
Ch
ina, 54
00–5405, 2006
.
[18]
F.M. Sun, X.S. Xu, and Z.H.
Qi, “Non-linear vibration and d
y
n
a
mic char
act
eris
tic of fish-lik
e r
obot controlled
b
y
GMM a
c
t
ua
tor”
,
Journal o
f
In
telligent Materia
l
S
y
stem and S
t
ructures
, Vol. 20, No. 12
, pp
. 1503–
1513, 2009
.
[19]
J
.
E
.
Colg
at
e a
nd K.M
.
L
y
nch
,
”M
echan
ic
s
and contro
l of
swimming: A review”,
IEEE Jo
urnal of Ocean
i
c
Engineering
, Vol. 29
, No
. 3
,
pp
.
660–673, 2004
.
[20]
M.
Sfa
k
iota
kis,
D.
M.
La
ne
, a
n
d J.
B.
C.
Da
vie
s
,
“
R
eview of fish
swimm
ing m
o
des for aqua
tic
lo
com
o
tion”,
IE
EE
Journal of Ocea
nic Engineering
, Vol. 24, No. 2
,
pp. 237–252
, 19
99.
Evaluation Warning : The document was created with Spire.PDF for Python.