IAES Inter
nat
iona
l
Journ
al
of A
r
tifici
al I
nt
el
li
gence
(I
J
-
AI
)
Vo
l.
8
, No
.
2
,
J
un
e
201
9
, pp.
95
~
106
IS
S
N: 22
52
-
8938
,
DOI: 10
.11
591/ijai.
v
8
.i
2
.pp
95
-
106
95
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/o
nline/i
nd
ex
.ph
p/I
J
AI
Develop
ment of
path pla
nn
ing algorithm
of
ce
ntip
ede inspi
re
d
wheeled
ro
b
ot
in
presenc
e of st
atic and mo
ving obs
tacles us
ing
modified
critic
al
-
snakebug
algorit
hm
Subir K
um
ar Da
s
1
, Aj
oy Ku
mar
Dutt
a
2
, S
ubir Ku
mar
Debnat
h
3
1
Depa
rtment
of C
om
pute
r
Appli
ca
t
ion
,
As
ansol Engi
ne
eri
ng
Co
l
le
ge
,
Ind
ia
2
,3
Depa
rtment
of
Producti
on
Enginee
ring
,
Jad
avpu
r
Univer
sit
y
,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Dec
20, 201
8
Re
vised
Ma
r
6
,
2019
Accepte
d
Ma
y
5,
2019
Path
pla
nn
ing
f
or
a
m
ovable
r
obot
in
r
eal
l
ife
situa
t
ion
h
as
bee
n
wi
d
e
l
y
cul
ti
v
ated
and
b
ec
om
e
rese
arc
h
int
er
est
for
la
s
t
few
de
ca
d
es.
Biom
imeti
c
robots
have
inc
r
ea
sed
a
tt
ra
ct
ion
for
the
ir
c
apa
b
ilit
y
to
dev
el
op
v
ari
ous
kind
of
walki
ng
in
orde
r
to
navi
g
a
te
in
different
envi
ronm
ent
.
T
o
m
ee
t
thi
s
re
quire
m
ent
of
nat
ura
l
insec
t
lo
comotion
has
ena
ble
d
th
e
developm
ent
of
compos
it
e
ti
n
y
r
obots.
Alm
ost
al
l
insec
t
-
sc
al
e
l
eg
ged
robots
ta
ke
m
oti
vat
ion
from
stiff
-
bod
y
hexa
pods;
thoug
h,
a
d
iffe
r
ent
dis
ti
nctive
org
ani
s
m
we
find
in
nat
ure
is c
en
ti
pe
de,
disti
nguis
h
ed
b
y
it
s num
ero
us l
egs
and
pli
abl
e bod
y
.
Thi
s
unique
ness
is
a
nti
ci
p
ated
to
pre
sent
per
form
ance
bene
f
it
s
to
bu
il
d
robot
o
f
the
said
t
y
p
e
in
te
rm
s
of
sw
ift
ness,
stea
dine
ss
,
toughne
ss
,
and
ada
pta
t
ion
abi
lit
y
.
Thi
s
p
ap
er
proposes
a
lo
ca
l
pat
h
pla
nn
in
g
al
gor
i
thm
of
m
ult
ipl
e
r
ake
ce
nt
ipe
de
inspire
d
robot
na
m
ely
Modifi
ed
Crit
ical
-
Snak
eB
ug
(MCS
B)
al
gorit
hm
.
Algo
rit
hm
tri
es
to
avoi
d
static
an
d
d
y
nami
c
obstacle
both
.
The
r
esult
s d
emons
tra
te t
he
c
apabili
t
y
of the
al
go
rit
hm
.
Ke
yw
or
d
s
:
Bi
oin
sp
i
red m
ob
il
e
rob
ot
Bug al
gorithm
Crit
ic
al
p
oin
t
Local
path pla
nn
i
ng
Ob
sta
cl
e a
void
ance
Op
e
n p
oin
t
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Subir
Ku
m
ar
Das
,
Dep
a
rtm
ent
of
Com
pu
te
r
A
pp
li
cat
ion
,
Asansol
En
gine
er
in
g
Coll
e
ge Asan
so
l,
West Ben
gal,
Ind
ia
.
Em
a
il
s: sub
ir
krdas@
gm
ai
l.com
1.
INTROD
U
CTION
Creat
ing
bio
lo
gical
ly
insp
ire
d
ti
ny,
res
ponsi
ve
it
ineran
t
rob
ots
has
al
ways
bee
n
gre
at
interest
in
researc
h
a
rea.
Asp
i
rati
on
f
or
rob
ots
to
trav
e
l
an
d
c
on
t
ro
l
obj
ect
s
fl
e
xib
ly
and
in
de
pende
ntly
in
3D
sit
ua
ti
on
,
reg
a
rd
le
ss
of
t
he
pr
ese
nce
,
a
bs
e
nce
or
dire
ct
ion
of
any
e
xter
nal
co
ntr
ol
syst
e
m
at
tract
s
resea
rchers.
This
insp
irat
io
n
is
he
lping
to
de
vel
op
va
rio
us
bi
om
i
m
et
ic
ro
bots
.
O
ne
of
th
e
re
aso
ns
to
beh
i
nd
this
is
eac
h
ki
nd
of
bio
lo
gical
crea
ture
ha
s
it
s
ow
n
ou
tst
a
nd
i
ng
dyna
m
is
m
c
apab
il
it
y
to
act
and
tra
vel
throu
gh
s
pecific
ty
pe
of
env
i
ronm
ent, inclu
ding c
riti
cal
an
d co
nf
i
ne
d place
. Som
e
of
t
he biolo
gic
al
insp
irat
io
ns
are ex
c
el
le
nt clim
ber
s,
so
m
e
wond
e
rful
s
wim
m
ers,
and
s
om
e
can
fly
,
so
m
e
jum
p
ver
y
well
.
T
he
nee
d
for
robo
ts
to
be
m
ob
il
e
in
stran
ge
an
d
diff
ic
ult
en
vir
on
m
ents
is
rap
id
ly
increasing.
This
incr
easi
ng
dem
and
br
in
gs
f
oc
us
on
di
ff
e
ren
t
kind
of
bio
l
ogic
al
ly
insp
ired
rob
ots.
Wh
il
e
discussi
ng
ab
ou
t
path
plan
ni
ng
t
he
pr
o
pe
rty
of
e
nv
i
ronm
ent
or
the
local
it
y
w
her
e
the
r
obot
will
wo
rk
is
an
i
m
po
rtan
t
issue.
If
th
e
env
ir
on
m
ent
is
sta
t
ic
then
al
l
the
env
ir
onm
e
ntal
el
e
m
ents
or
obsta
cl
es
ar
e
sta
ti
c.
In
case
of
dynam
ic
env
i
ronm
ent
t
he
el
e
m
ents
m
ay
be
sta
ti
c
or
m
a
y
be
dy
nam
ic
.
T
he
r
obot
m
ay
and
m
ay
no
t
know
al
l
the
inf
or
m
at
ion
ab
out
the
env
ir
onm
ent.
In
dynam
ic
env
ir
on
m
ent
,
path
plan
ning
com
es
in
two
form
a)
Global
pat
h
plan
ning
–
w
her
e
th
e
inf
orm
at
ion
about the envir
on
m
ent is k
nown
to
r
obot b
e
fore it st
arts
mo
vi
ng
a
nd
b)
L
ocal p
a
th p
la
nning
–
w
he
re the
r
obot
do
e
sn’t hav
e a
ny infor
m
at
ion
r
egardin
g
the
env
i
ronm
ent. Th
is i
s sen
sory
b
ased pat
h
pla
nn
i
ng
wh
e
re the
r
obot
us
es
se
nsors
t
o
get
in
f
or
m
ation
a
bout
the
ob
sta
cl
es
to
a
void
it
an
d
reac
h
desti
nation.
In
case
of
loc
al
path
plan
ning
rob
ot
acq
uires
i
nform
at
ion
ab
ou
t
l
ocali
ty
durin
g
it
s
m
ov
em
ent.
In
unkn
own
dy
nam
ic
env
ir
onm
ent,
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8938
IJ
-
AI
V
ol.
8
,
No.
2, June
201
9
:
95
–
106
96
nav
i
gation
al
gorithm
of
m
ob
il
e
ro
bot
av
oidi
ng
obsta
cl
e
can
be
done
us
i
ng
f
uture
pre
dicti
on
with
t
he
he
lp
of
pr
i
or
it
y
be
hav
i
or
[
1].
The
sai
d
process
us
es
sing
le
r
a
ke
r
obot.
But
in
cas
e
of
bi
ologica
ll
y
insp
ire
d
r
ob
ot
li
ke
sn
a
ke
r
obot
th
e
nav
i
gatio
n
r
equ
i
res
a
diff
e
ren
t
kind
of
lo
com
otion
te
ch
niques
a
nd
m
echan
ic
s
[
2].
A
sn
a
ke
rob
ot
sh
ould
be
so
desig
ne
d
and
m
od
el
ed
physi
cal
ly
as
well
m
a
the
m
at
icall
y
so
that
it
beh
ave
s
li
ke
rea
l
sn
ake
[3
]
. T
he
m
otion
p
at
te
r
n
of snake li
ke
r
obot
sh
oul
d
be
c
on
t
ro
ll
ed
in s
uc
h way t
hat it
w
il
l
b
e able to
nav
i
gate on
la
nd
as
well
in
water
[4
]
.
T
he
m
et
ho
d
t
o
co
nt
ro
l
m
ov
em
ent
of
Sn
a
ke
rob
ot
shou
l
d
be
well
def
i
ned
[
5
-
7].
This
kind
of
rob
ot
m
ay
us
e
wh
eel
s
for
m
otion
.
Othe
r
interest
i
ng
bio
l
ogic
al
ly
insp
ir
ed
rob
ot
is
centi
pe
de
-
li
ke
m
illi
rob
ot [8]. T
hes
e k
in
ds
of ro
bot’s g
ai
ts a
re
now on
e
of the
m
os
t i
ntere
sti
ng
researc
h
t
op
ic
s
. The n
a
vi
gatio
n
a
nd
ob
sta
cl
e
a
voidi
ng
pat
h
plan
ni
ng
te
ch
nique
of
a
m
ob
i
le
r
ob
ot
with
m
ulti
pl
e
ra
kes
us
in
g
vi
su
al
or
oth
e
r
s
ens
or
s
therefo
re
pay
at
te
ntion
m
or
e
and
m
or
e
[9
-
17
]
.
A
par
t
f
r
om
sn
ake
a
nd
centi
pede
ot
he
r
ki
nd
s
of
stu
died
bio
m
i
m
et
ic
r
obots ar
e
sc
orpio
n [18,
19] an
d
i
nch wo
rm
[
20
]
.
Bug
al
gorithm
s
are
one
of
the
us
e
d
s
i
m
ple
local
path
plan
ni
ng
te
chn
i
qu
e
.
Crit
ic
al
-
Po
i
ntBu
g
al
gorithm
[2
1]
belo
ngs
t
o
bug
al
go
rithm
race
that
propos
es
a
procedu
re
to
reach
de
sti
nation
av
oi
ding
sta
ti
c
ob
sta
cl
es.
Howev
e
r
this
al
gorithm
con
sid
ers
a
po
i
nt
robo
t.
A
de
velo
pm
ent
fr
om
t
he
sai
d
al
gori
thm
is
Crit
ic
al
-
Sn
ake
Bug
al
gorithm
[22].
This
al
gorithm
m
od
ifie
s
the
Crit
ic
a
l
-
Po
i
ntBug
al
gorithm
so
that
the
m
od
ifie
d
ve
rsion
ca
n
be
use
d
for
bi
o
-
i
ns
pi
re
d
sna
ke
li
ke
m
ulti
ple
rak
e
r
obot.
The
al
gori
thm
exists
in
[2
2]
is
us
ef
ul
f
or
sta
ti
c
ob
sta
cl
e
av
oid
ance
durin
g
pa
th
plan
ning
f
or
bio
-
ins
pire
d
m
ul
ti
ple
rak
e
r
obot
w
her
e
eac
h
ra
ke
is
assum
ed
to b
e
a point.
B
ut
obsta
cl
e
m
a
y
no
t be
al
ways stat
ic
and
th
e
r
akes o
f
the r
ob
ot
m
ay
hav
e
som
e
non
neg
li
gi
ble val
ue
in
it
s d
im
ension. In
t
hose ca
ses the al
gorith
m
ex
ist
s i
n
[22
]
m
ay
n
ot work
w
el
l.
T
he pr
op
os
e
d
Mod
ifie
dCriti
cal
-
Sn
a
keBu
g(
MC
SB)
al
gorit
hm
her
e
is
a
m
od
ific
at
ion
a
nd
de
velo
pm
e
nt
of
al
gorith
m
pr
esent
in
[
22
]
f
or
m
ulti
ple
rak
e
ce
ntipede
i
nspired
r
obot
to
a
vo
i
d
s
ta
ti
c
as
well
as
dynam
ic
ob
sta
cl
e.
T
h
e
al
go
r
it
h
m
pro
po
se
d
he
re
con
side
rs
the
rak
es
of
rob
ot
to
hav
e
di
m
ension
not
as
po
int.
T
her
e
fore
durin
g
obsta
cl
e
avo
i
dan
ce
it
w
i
ll
calc
ulate
the
safety
precauti
on also
an
d
t
hus the
ge
ner
at
e
d path
is m
or
e r
eal
ist
ic
.
2.
PROP
OSE
D MET
HO
D
Im
agine
a
bu
g
is
m
ov
ing
to
a certai
n
d
irect
io
n
an
d
faces w
it
h
an
obstr
uctio
n
in
it
s
path.
T
o
av
oid
th
e
ob
st
ru
ct
io
n
the
bug
ci
rcu
m
nav
igate
s
it
till
t
he
m
otion
to
t
he
or
i
gin
al
pat
h
is
no
t
furthe
r
blo
c
ked.
The
bug
al
gorithm
s
fo
r
local
path
pla
nnin
g
a
nd
obsta
cl
e
avo
i
ding
ar
e
insp
i
red
by
t
his
be
ha
v
io
r
of
natu
ral
bu
g.
T
hese
are
sens
or
de
pe
nd
e
nt
pat
h
pl
ann
i
ng.
Tw
o
m
ai
n
beh
a
vio
r
s
of
these
al
go
rithm
s
are:
1.
m
ov
ing
to
the
obj
ect
ive
and
2.
bounda
r
y
fo
ll
owin
g
of
an
e
nc
ountere
d
ob
sta
cl
e.
Bu
g1,
B
ug2
wer
e
the
pr
im
ary
al
gorithm
s
introdu
c
e
d
by
L
um
el
sk
y
et
al
[
21]
.
Si
nc
e
then
a
nu
m
ber
of
ver
si
ons
ha
ve
bee
n
ra
ise
d
of
bug
al
gorithm
s
each
with
a
dev
el
op
m
ent
than
t
he
pre
vious
one.
M
os
tl
y
us
ed
a
nd
note
d
in
path
plan
ni
ng
of
m
ob
il
e
r
obot
are
Ta
nge
ntBu
g,
VisBu
g,
DistB
ug
et
c.
Exce
pt
these
few
ot
her
al
gorithm
s
are
Re
v,
Re
v2,
O
neBu
g,
L
eaveBu
g,
P
oint
Bug
,
K
-
B
ug,
Crit
ic
al
-
Po
i
ntBug,
Cr
it
ic
a
-
Sn
akeB
ug
et
c.
T
he
m
ai
n
ob
j
ect
ives
of
eve
ry
bu
g
al
gorithm
s
at
tem
pt
to
sh
ort
en
the
path,
sho
rten
the
tim
e
to
reach
de
sti
nation,
sim
plify
the
al
go
ri
thm
and
m
aking
it
m
o
re
reali
s
ti
c
in
ever
y c
o
nd
it
io
n,
s
how
n
i
n
F
i
gure
1.
Figure
1.
Tra
j
e
ct
or
ie
s
form
ed
by d
i
ff
e
ren
t
bu
g
al
go
rithm
with r
e
fer
e
nce t
o l
it
eratur
e [
20]
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
IS
S
N:
22
52
-
8938
Develo
pm
e
nt
of
p
at
h pla
nn
i
ng a
lg
or
it
hm
of c
entipede
ins
pir
ed
w
heele
d
r
ob
ot
...
(
Subir K
umar
Das
)
97
The
C
riti
cal
-
Sn
akeB
ug
al
go
r
it
h
m
[22]
is
de
velo
ped
keepi
ng
i
n
m
ind
of
m
ulti
rak
e
r
obot.
This
al
gorithm
trie
s
to
avo
i
d
sta
ti
c
ob
sta
cl
e
of
al
l
it
s
li
nk
s
wh
il
e
m
ov
in
g
from
a
so
urce
point
to
destinat
io
n
po
i
nt.
Using
it
s
body
gest
ur
e
na
viga
ti
ng
th
r
ough
cl
uttered
a
nd
r
ough
s
urr
ound
ing
s
is
m
uch
easi
er
f
or
s
nake
li
ke
m
ul
ti
rak
e
r
obot
than
ot
her
robot.
Li
ke
ot
he
r
m
os
tl
y
us
ed
bug
al
go
rithm
s
this
al
gorithm
al
so
us
e
s
dm
in
li
ne
to
get
the
m
ini
m
um
distance
f
ro
m
robo
t
to
destinat
io
n.
T
he
dm
in
help
s
the
r
obot
to
reach
destinat
i
on
by
traversi
ng
m
inim
u
m
distance.
The
m
ajo
r
ad
van
ta
ges
of
Bu
g
al
gorithm
s
ov
er
the
oth
er
e
xisti
ng
path
pl
ann
i
ng
and
obsta
cl
e
avo
i
ding
al
go
r
it
h
m
is
it
s
sim
pl
ic
it
y
and
us
e
of
m
ini
mu
m
nu
m
ber
s
ens
or
s.
T
his
pro
posed
al
gorithm
us
es
the
te
chn
i
qu
e
of
on
e
m
or
e
al
gorithm
na
m
ely
“Op
er
at
iveCrit
ic
al
Po
int
Bu
g”,
wh
ic
h
is
ye
t
to
be
publishe
d.
T
he
al
go
rithm
c
on
si
ders
dyna
m
ic
path
plann
in
g
kee
ping
i
n
m
ind
of
th
e
di
m
ension
of
rob
ot
.
Th
ough it
is no
t y
et
p
ubli
sh
e
d so no
t
r
e
fen
ce
d i
n
li
te
ratu
re r
e
view.
2.1.
M
od
i
fied
cri
tical
–
sn
ak
e
bug a
l
go
ri
th
m
This
al
gorithm
assist
s
to
locat
e
the
path
and
pass
thr
ough
it
fo
r
a
bio
l
og
ic
al
ly
centi
pe
de
ins
pire
d
wh
eel
e
d
m
o
bile
rob
ot
in
a
pl
ane
fl
oor
fill
ed
wit
h
obsta
cl
es.
T
he
obsta
c
le
s
are
unide
nt
ifie
d
an
d
of
sta
nd
a
r
d
form
,
di
m
ension,
a
nd
locat
io
n.
To
reac
h
th
e
ta
rg
et
from
current
point
it
cal
culat
es
an
d
determ
ines
th
e
ne
xt
po
i
nt
to
m
ov
e
by
getti
ng
on
t
he
ob
sta
cl
e
loc
at
ion
.
Fig
ure
2
shows
an
exa
m
ple
of
centi
pe
de
li
ke
18
w
he
el
e
d
m
ob
il
e
ro
bot
[
23
]
known
a
s
Tam
iy
a
70
230.
This
al
go
rith
m
helps
to
na
vi
gate
a
r
obot
in
pla
nar
of
unknown
locat
ion
w
hich
is
fill
ed
with
s
ta
ti
c
as
well
as
dynam
ic
ob
sta
cl
es
of
unde
fin
ed
but
regular
sh
a
pe.
Ra
nge
sen
s
or
is
us
ed
to
rec
ognize
a
n
al
te
ra
ti
on
in
distanc
e
to
recog
nise
ob
sta
cl
e
posit
ion
s
.
Fig
ure
3
sh
ows
t
he
sca
nn
i
n
g
process
of
a
r
ang
e
se
ns
or
i
n
it
s
fiel
d
of
view
(FO
V)
t
o
detect
obsta
cl
es
[21].
We
belie
ve
pe
rh
a
ps
a
n
unboun
ded
s
pa
ce
Q
⊂
R
2
occu
pie
d
by
a
set
of
bounde
d
sta
ti
c
and
dynam
i
c
ob
sta
cl
es
O
=
{O
1
,O
2
,
.
.
.
,
O
K
}.
The
w
hole
en
vir
on
m
ent
is
c
on
si
der
e
d
to
be
in
2D
with
t
he
obj
ect
i
ve
of
si
m
ulati
on
.
F
ur
t
her
we
c
ons
ider
a
rob
ot
w
hich
is
m
ade
up
of
a
series
of
w
he
el
ed
r
obot.
Th
e
first
wh
eel
e
d
rob
ot
ra
ke
is
furn
is
he
d
with
rang
e
sens
or
t
o
se
nse
obsta
cl
es.
Fi
g
ure
4
il
lustrat
es
the
head
of
the
centi
ped
e
insp
ire
d
rob
ot
in
rectan
gula
r
sh
a
pe
with
w
heel
a
nd
ot
her
e
quip
m
ent.
The
othe
r
ra
kes
c
on
ta
i
n
a
com
pact,
m
ul
ti
-
cor
e
pro
pelle
r
m
ic
ro
co
ntr
oller
with
ha
r
dw
a
re
to
com
m
un
ic
at
e
betwee
n
eac
h.
We
pu
t
th
e
rob
ot
an
d
the
ob
sta
cl
e
in
2D
plane
a
nd
ass
um
e
a
virtu
al
ci
rcle
s
urrou
nd
i
ng
the
r
obot.
E
ve
ry
par
t
of
the
r
obot
will
be
c
onsidere
d
in
side
a
virt
ual
ci
rcle
.
T
he
head
a
nd
oth
er
r
akes ar
e sam
e in al
m
os
t a
ll
aspect b
ut the o
nly dif
fer
e
nce
is a sen
so
r
is at
ta
ched
w
it
h
th
e h
ead
wh
ic
h
can
sca
n
180
0
a
rea.
D
ur
i
ng
m
ov
em
e
nt
ever
y
ra
ke
or
sect
ion
of
the
robo
t
s
uppose
d
to
go
after
t
he
sa
m
e
coor
din
at
es
of
it
s
pr
edece
ssor.
T
he
init
ia
l
and
final
c
oord
i
nate
with
respec
t
to
global
f
ram
e
of
ref
e
re
nce
is
known
to
r
obot
.
Pr
i
or
we
pr
oc
eed
t
o
the
des
cripti
on
of
al
gorithm
s,
we
co
m
po
se
so
m
e
im
po
rtant
an
d
help
fu
l
assum
ption
a
nd
def
i
niti
on
f
or this alg
or
it
hm
wh
ic
h
a
re alm
os
t sam
e as [
22
]
.
Fi
gure
2.
Ce
ntipede
li
ke wh
e
e
le
d
rob
ot w
it
h re
fe
ren
ce
to [2
3]
Figure
3.
O
bs
t
acl
es d
et
ect
ed by ra
nge
sens
or(R)
Figure
4.
Dis
play
o
f
robot a
nd
ob
sta
cl
es i
n 2
D plot
2.1.1.
As
sum
pt
ion
The ro
bot i
s s
uppose
d be as
f
e
w ph
ysi
cal
ly
an
d analy
ti
cal
ly conn
ect
e
d
sm
al
l robo
ts
The
al
gorithm
u
ses
w
or
ld
c
o
-
ordinate syst
e
m
as g
lob
al
fra
m
e o
f refe
renc
e
All p
os
it
ion co
ordinates
(incl
ud
i
ng w
it
h i
nit
ia
l and
goal
)
a
r
e at first
quad
r
ant
The vel
ocity
a
nd ang
ular vel
ocity
o
f
robot i
s constant i
n
e
ver
y m
ov
em
ent an
d r
otati
on r
especti
vely
The
e
nviro
nm
ent is flat
a
nd m
ai
ntains e
qu
al
al
ti
tud
e eve
rywh
e
re
The
m
ob
il
e r
obot m
ov
es
in
a
two
-
dim
ensional
sp
ace a
nd ro
ta
te
s w
it
hout sl
ipp
in
g
Both
the
rob
ot
and
dynam
ic
ob
sta
cl
es
a
re
r
un
i
n
co
ns
ta
nt
sp
ee
d
an
d
in
st
raig
ht
li
ne.
I
f
anyo
ne
wa
nts
t
o
change
directi
on it
h
a
s to
sto
p t
hen
t
urn
t
hen
again st
art m
ov
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8938
IJ
-
AI
V
ol.
8
,
No.
2, June
201
9
:
95
–
106
98
2.1.2.
Imp
or
t
an
t
de
finitio
ns
The
r
obot
sca
ns
t
he
s
urr
oundin
gs
0
0
t
o
180
0
by
ra
nge
se
ns
or.
S
o
t
o
m
a
ke
a
sca
n
of
total
36
0
0
t
he
rob
ot
has
to
r
ot
at
e
the
head
pa
rt
90
0
le
ft
an
d
righ
t
a
nd
s
ca
n
the
env
i
ronm
e
nt
again
.
O
bv
i
ou
sly
it
is
req
ui
red
i
f
the
centi
pe
de
ge
ts
fu
ll
of
obst
acl
e
at
18
0
0
sc
an.
We
co
ns
i
de
r
on
bo
unde
d
and
unbo
unde
d
sp
ac
e
O
a
nd
Q
in
a
read
i
ng
from
ran
ge
sens
or
wi
thin
a
ti
m
e
per
iod
,
t
n
to
t
n+1
.
I
f
it
no
ti
ces
a
va
riat
ion
in
dist
an
ce
say
Δd
(
wh
e
re
Δd
value
is
defi
ned)
in
range at
any
po
sit
ion
then
that
posit
ion
point
is
regarde
d
as
O
pen
Po
int.
T
he
pri
m
aril
y
the r
obot f
aci
ng d
irect
ly
to
goal
p
oin
t an
d
aft
erw
a
rd
it
b
egi
ns scann
i
ng
fo
r
op
e
n
point. S
ub goal po
i
nt is
a
m
id
obj
ect
ive
of
t
he
r
obot
t
o
at
ta
in
the
fi
nal
obj
ect
ive
.
T
his
point
is
placed
at
a
par
ti
cular
differe
nc
e
from
the
co
rr
es
pond
ing
op
e
n
po
i
nt
and
perha
ps
create
s
a
rig
ht
ang
le
at
ope
n
po
i
nt
to
the
l
ine
passi
ng
t
hroug
h
the
se
ns
or
poin
t
of
r
obot
a
nd
t
he
open
point.
Fr
om
the
c
oor
din
at
e
of
op
e
n
po
i
nt
c
orrespo
nd
i
ng
s
ub
goal
point
is
determ
ined.
The
r
obot
sel
e
ct
s
on
e
po
i
nt
cal
le
d
crit
ic
al
po
int,
f
or
subse
qu
e
nt
m
ov
e
from
a
set
of
su
b
goa
l
po
i
nts.
T
his
po
int
is
sel
ect
ed
on
t
he
ba
se
of
sm
a
ll
est
distance
from
the
ta
rg
et
point
an
d
s
o
f
a
r
not
co
ns
i
der
e
d
for
m
ov
em
ent.
The
r
obot
at
te
m
pts
to
m
ov
e
i
ts
each
rak
e
’s
center
of
vi
rtu
al
ci
rcle
to
c
riti
cal
po
i
nt
or
t
o
virtu
al
ci
rcle’s
center
of
it
s
pr
e
deces
so
r
.
The
m
idp
oi
nt
of
the
tw
o
e
xtrem
e
po
ints
of
eac
h
r
obot’s
rak
e
is
centre
of
t
he
corres
pondi
ng
virtu
al
ci
rcle
a
nd
the
half
dis
ta
nce
bet
ween
the
tw
o
e
xtre
m
e
po
ints
of
t
he
ra
ke
with
a
safety
const
ant
is
the
ra
diu
s
.
The
lowest
valu
e
of
ca
n
be
ob
ta
ined
f
ro
m
few
num
ber
of
e
xp
e
rim
ents.
Fig
ure
4
sh
ows
the
plo
tt
ing
a
nd
oth
e
r
r
equ
i
red ge
om
e
tric
al
d
et
a
il
s.
d
–
Per
pendicular
distance
of obs
ta
cl
e fr
om
the
sens
or
A
–
Locati
on
of
r
obot
w
heel
-
‘a
xl
e’ (
im
aginar
y l
ine)
ce
nter
X
–
Dista
nce
from
sen
s
or
t
o o
pen
po
i
nt
w
–
Dista
nce
betwe
en wh
eel
-
‘a
xle
’
ce
nter
a
nd wheel
cente
r
F
–
Dista
nce
betwe
en
the
r
ea
r
e
nd
and
w
heel
‘a
xle’
ce
nter
R
–
Dista
nce
betwe
en
the
wheel
‘a
xle’
ce
nter
and
the
rear
e
nd
of the
rob
ot
C
–
Ce
nter of
the
vi
rtual circl
e
α
–
Op
e
n p
oin
t
detect
ion
a
ng
le
b
y
the se
nsor
-
Safety
constant
for
t
he ro
bot
r
–
Ra
dious
of the
virtu
al
ci
rcle
β
–
Sensor
d
i
recti
on a
ng
le
at
ope
n p
oin
t
with
res
pect to t
he
li
ne
p
a
rall
el
to
x
-
a
xis a
nd p
as
sin
g t
hro
ugh (x
i
,y
i
)
befor
e
m
ov
em
ent
θ
–
Angle
gen
e
rate
d by β
with
res
pect to t
he
li
ne
p
a
rall
el
to
x
-
a
xis fo
r
s
ub goa
l po
i
nt coo
rd
i
na
te
s calc
ulati
on
–
Angle c
reated
on a line
pa
rall
el
to
x axis
and
p
assi
ng th
rou
gh se
nsor by a
li
ne fr
om
sen
sor
to op
e
n p
oin
t
–
Angle
betwee
n l
ine joini
ng
open po
i
nt & su
bg
oal a
nd
the
li
ne parall
el
to x
-
axis a
nd
passi
ng th
rou
gh
op
e
n p
oin
t
d
k
–
Dista
nce
of a s
ub goal
po
i
nt fro
m
cu
r
ren
t l
oc
at
ion
x
s
a
nd y
s
–
a
re
the sig
n fact
ors
u
se
d
i
n determ
inati
on
of the
c
oor
din
at
es
of open
points
s
x
an
d
s
y
–
are t
he
si
gn f
act
ors
us
e
d
to
d
et
erm
ine the c
oor
dina
te
s o
f
s
ub goa
l po
i
nts
We c
onsider
,
T
n
= {(
x
1
,y
1
),(x
2
,y
2
),
...,(
x
i
,y
i
)}
as a set
po
i
nts
t
rav
el
le
d
by t
he
robot
’s
n
t
h
li
nk
wh
e
re
(x
i
,y
i
) sym
bo
li
ze the
coor
din
at
e
values
t
he nth lin
k arri
ved
OP
=
{
(
(α
a
,d
a
)
,
(α
b
,d
b
)),....,
((
α
k
,d
k
)
,(
α
j
,d
j
)
)
}
as
a set o
f
op
e
n
points
of obstac
l
es ide
ntifie
d
by
the se
ns
or
wh
e
re α a
nd
d i
nd
ic
at
e the
angle
s
&
d
ist
an
c
e
s of
op
e
n p
oint
s
res
pecti
vely
f
r
om
the r
ob
ot
and each
(
(α
a
,d
a
),(α
b
,d
b
)) re
pr
ese
nts tw
o o
pen points
conn
ect
e
d
to
eac
h ob
sta
cl
e,
if
only
one
po
i
nt i
s d
et
ect
ed
then
α
a
=
α
b
a
nd
d
a
=
d
b
SG
=
{
(a
i
,b
i
)
,
…
..,(x
j
,y
j
)
}
as a
s
et
o
f
sub g
oal
po
i
nts r
ec
ogniz
ed
by
the
r
obot
an
d
e
ver
y
(a
i
,b
i
)
sig
nifies t
he
su
b g
oal
point
wh
ic
h
the
ro
bot can selec
t
for next m
ov
e
T
obs
= {((
a
i
,b
i
)
,
…(a
j
,
b
j
)
),
…
..,(
(x
i
,y
i
)
,..,(
x
j
,y
j
)
)
} as a set
of
ini
ti
al
l
y
identifie
d
obstr
uction w
her
e
each
((a
i
,b
i
)
,…
(a
j
,b
j
)
)
are
th
e
set
of
po
i
nts
of
eac
h
ob
sta
cl
e.
S
obs
= {((
a
i
,b
i
),…(a
j
,b
j
)
),
…
..,(
(x
i
,y
i
)
,..,(
x
j
,y
j
)
)
} as a set
of f
i
xe
d obst
acl
es where
ev
e
ry ((a
i
,b
i
),…(a
j
,b
j
))
a
r
e
the set
of
point
s
of eac
h fixe
d ob
sta
cl
e.
TD
obs
= {(
(a
i
,b
i
)
,…(
a
j
,b
j
))
,
…..,(
(x
i
,y
i
)
,..,(
x
j
,y
j
)
)
} as a set
of
ini
ti
al
ly
m
ov
in
g
obsta
cl
es whe
re
ever
y
((a
i
,b
i
)
,…
(a
j
,b
j
)
)
are
the
set
of
po
i
nts
of
eac
h pro
vision
al
ly
m
ov
ing
obsta
cl
e.
D
obs
= {(
(a
i
,b
i
),…(a
j
,b
j
))
,…
..,(
(x
i
,y
i
)
,..,(
x
j
,y
j
))}
as a set
of m
ov
a
ble
obsta
cl
es where
each
((
a
i
,b
i
)
,…
(a
j
,b
j
)
)
are the
s
et
of
points
of each
m
ov
a
ble
obsta
cl
e.
D= {((
x
a
,y
a
),δ
a
)
,…
,
(
(
x
j
,y
j
)
,δ
j
)
} as a set
of s
ub
go
al
points a
nd d
ist
a
nce fr
om
ta
rg
et
of t
hat point
wh
e
re ea
ch
set
(
(x
a
,y
a
)
,δ
a
)
r
epr
ese
nts t
he
s
et
o
f
sub g
oal
po
i
nt and t
heir dist
ance
from
t
arg
et
Her
e
d
m
in
is t
he
d
ist
ance
fro
m
the ro
bo
t t
o goal
p
oi
nt and
i
s the
directi
on
of the sam
e.
P
C
is t
he p
os
it
ion
w
he
re th
e
r
obot
m
ay
co
ll
i
de wit
h
the
obs
ta
cl
e
and
the
di
sta
nce
f
r
om
cu
r
re
nt positi
on
of
rob
ot to
t
he po
i
nt of c
olli
sion i
s
D
C
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
IS
S
N:
22
52
-
8938
Develo
pm
e
nt
of
p
at
h pla
nn
i
ng a
lg
or
it
hm
of c
entipede
ins
pir
ed
w
heele
d
r
ob
ot
...
(
Subir K
umar
Das
)
99
2.1.3.
Al
go
ri
t
hm
a.
Ma
in pro
ce
dur
e
1.
Robot Start
2.
Take i
nput
of the
posit
ion
c
o
-
ordinates
of s
ource
and
destin
at
ion
3.
Ca
lc
ulate
the dist
ance and
dir
ect
ion
from
so
ur
ce
to desti
nat
ion
d
m
in
and
resp
ect
ively
4.
W
H
ILE
not
D
est
inati
on
5.
Start OBS
TAC
LE_
DETECT
I
ON proc
ed
ur
e
in F
OV
6.
IF
obsta
cl
e or
virtu
al
obsta
cl
e in
directi
on
7.
Ca
lc
ulate
the
c
oor
din
at
es
of
s
ub
goal
s
from
OP
,
do
n’
t
cal
c
ulate
sam
e
set
from
OP
twic
e
an
d
save it i
n set
S
G
a
nd D
8.
Ca
lc
ulate
d
ist
ance
of each
sub g
oal
from
d
est
inati
on
a
nd s
ave it i
n set
D
9.
Sele
ct
the c
oor
din
at
e
po
i
nt P
hav
i
ng the l
owest
d
ist
ance i
n D
10.
IF
t
he po
i
nt ex
i
sts i
n
T
rav
e
rse po
i
nt set T
1
11.
Discar
d
the
poi
nt
12.
Sele
ct
the
nex
t
lowest
distance
point P
fro
m
D
13.
Fo
ll
ow ste
p 10
14.
ELSE
15.
IF
P
is S
AF
E
_POI
NT
16.
Sa
ve
t
he
c
oor
di
nate in tra
ve
rs
e point set
T
1
17.
Ca
lc
ulate
an
gl
e of rotat
ion a
nd
ro
ta
te
18.
Ca
ll
MOV
E
Proced
ure to
m
ov
e at P
19.
IF
obsta
cl
e d
y
nam
ic
20.
IF
VI
RT
U
AL
OBST
ACLE
21.
Fo
ll
ow ste
p 7
22.
END I
F
23.
END I
F
24.
Ca
lc
ulate
d
irec
ti
on
and
distance
d
m
in
25.
ENDIF
26.
ENDIF
27.
ELSE
28.
Ca
lc
ulate
the ne
xt point c
oor
di
nate P towa
rds the
directi
on
29.
IF
P
is S
AF
E
_POI
NT
30.
Save
t
he
c
oor
di
nate in T
1
31.
Ca
ll
MOV
E
Proced
ure to
m
ov
e at P
32.
Ca
lc
ulate
n
ew
d
m
in
33.
END I
F
34.
END I
F
35.
END
WHILE
36.
Robot St
op
b.
Mov
e
pro
ce
dur
e
1.
FO
R i
=
n
to
2
2.
Save
t
he
la
st c
oor
din
at
e
of
T
i
-
1
in tra
ver
se
po
int set
T
i
3.
END F
OR
4.
FO
R i
=
1
to
n
5.
Mov
e
the ith
li
nk to
ward t
he
l
ast
point
of
T
i
6.
END F
OR
c.
Ob
sta
cl
e
_d
et
ec
ti
on
proce
dure
1.
Scan 0
0
t
o 180
0
2.
Id
e
ntify o
bs
ta
c
le
s in
F
O
V
3.
Ca
lc
ulate
the c
oor
din
at
es
of open
points
and
save t
ho
se
poi
nts in
set T
obs
4.
IF
no ope
n p
oin
ts
5.
Ro
ta
te
the
hea
d
le
ft
90
0
6.
Scan 0
0
t
o 180
0
7.
Ca
lc
ulate
the c
oor
din
at
es
of open
points
and
save t
ho
se
poi
nts in
set T
obs
8.
Rotat
e the
hea
d rig
ht 90
0
9.
Scan 0
0
t
o 180
0
10.
Ca
lc
ulate
the c
oor
din
at
es
of open
points
and
save t
ho
se
poi
nts in
set T
obs
11.
END I
F
12.
FO
R a
ny c
oor
d
inate
in
T
obs
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8938
IJ
-
AI
V
ol.
8
,
No.
2, June
201
9
:
95
–
106
100
13.
IF
c
oor
din
at
e s
et
o
f
T
obs
exists
in
S
obs
14.
Discar
d
the
coor
din
at
e
from
T
obs
15.
ELSE
16.
IF
t
her
e e
xists
any n
ea
rest
po
i
nt in S
obs
with
sm
a
ll
ch
ang
e
in x
or y val
ue
17.
The o
bs
ta
cl
e m
ay
b
e
dynam
ic
18.
Save
c
oor
din
at
e in T
D
obs
19.
ELSE
20.
IF
t
her
e e
xists
any n
ea
rest
po
i
nt in TD
obs
with sm
al
l chan
ge
in x
or y
va
lue
21.
It is p
a
rt
of
m
ov
in
g obst
acl
e
22.
Save
t
he po
i
nts
of T
D
obs
and T
obs
to the ro
bo
t
reg
ist
ry a
nd all
the
op
e
n p
oin
t
s of
the
ob
sta
cl
e
in D
obs
23.
ELSE
24.
Save
t
he po
i
nts
in
S
obs
25.
RETURN
obst
acl
e
26.
END I
F
27.
END I
F
28.
END I
F
29.
END F
OR
30.
IF
obsta
cl
e d
y
nam
ic
31.
IF
VI
RT
U
AL
OBSTCLE
32.
RETURN
v
i
rtu
al
o
bst
acl
e
33.
END I
F
34.
END I
F
d.
Virtual
ob
sta
cl
e
pro
ce
dure
1.
Ca
lc
ulate
d
irec
ti
on
a
nd s
pee
d v
obs
of the
ob
sta
cl
e from
the points e
xists i
n rob
ot r
e
gistr
y
2.
IF
t
he ob
sta
cl
e
directi
on
i
ntersecti
ng ro
bo
t
directi
on
3.
Ca
lc
ulate
p
r
ob
able posit
io
n
P
C
of colli
sio
n
4.
Ca
lc
ulate
p
r
ob
able d
ist
a
nce
D
C
f
ro
m
the ro
bo
t t
o
P
C
5.
IF
D
C
<d
m
in
6.
Ca
lc
ulate
the
tim
e
R
T
C1
and
R
TC2
the
rob
ot
will
ta
ke
t
o
ar
rive
at
P
C
an
d
cr
os
s
P
C
fu
ll
y
resp
ect
ively
7.
Ca
lc
ulate
the
ti
m
e O
TC
m
ov
ing
ob
sta
cl
e
will
r
eq
uires
to
a
rr
i
ve
at
P
C
8.
IF
R
TC2
>=O
TC
and R
TC1
<=O
TC
9.
Id
e
ntify t
he p
oi
nt P
C
an
d al
l t
he
adj
ace
nt
points
of the
obstacl
e
as
virtu
al
O
bs
ta
cl
e
10.
Ca
lc
ulate
the open points
and
save i
n OP
11.
RETURN
v
i
rtu
al
o
bst
acl
e
12.
ENDIF
13.
END I
F
14.
END I
F
e.
Safe
_poin
t
pro
cedure
1.
IF
P
corr
esp
on
ds
t
o
a s
ub s
ub
set
o
f
OP
2.
Sele
ct
the s
ub
set
f
r
om
O
P th
at
co
r
res
ponds
to P
3.
IF
it
is
first s
ub set
4.
Sele
ct
the sec
ond su
b set
from
p
rev
io
us
set
from
OP
5.
ELSE
6.
Sele
ct
the f
i
rst
su
b set
from
n
ext set f
r
om
O
P
7.
END
IF
8.
Ca
lc
ulate
the dist
ance S
d between
tw
o
s
ub s
et
s
9.
ELSE
10.
Sele
ct
the tw
o po
i
nts fr
om
O
P h
a
ving m
inim
u
m
d
ist
ance f
r
om
P
11.
Ca
lc
ulate
the dist
ance S
d between
tw
o po
i
nt
s
12.
END I
F
13.
IF
S
d
<
2r
14.
RETURN
saf
e
15.
ELSE
16.
RETURN
not
safe
17.
ENDIF
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
IS
S
N:
22
52
-
8938
Develo
pm
e
nt
of
p
at
h pla
nn
i
ng a
lg
or
it
hm
of c
entipede
ins
pir
ed
w
heele
d
r
ob
ot
...
(
Subir K
umar
Das
)
101
3.
DISCU
SSI
ON
A
ND R
ES
UL
T
The
pri
nci
ple
m
ot
ive
of
the
a
lgorit
hm
is
to
c
reate
a
con
ti
nu
ou
s p
at
h
f
r
om
fixe
d
source
to
destinat
ion.
Avoidi
ng
sta
ti
c
obsta
cl
e
can
be
is
discuss
e
d
in
[
20
-
21
]
,
the
Fig
ure
1
s
how
s
the
a
vo
i
dan
c
e
as
pe
r
[20].
Curren
t
al
gorithm
a
tt
e
m
pts
to
avo
i
d
m
ov
ing
obsta
c
le
too
.
Als
o
thi
s
al
go
rit
hm
in
con
t
rast
of
[
21
]
,
j
udges
t
he
r
obot
as
a
series
of
ra
kes
hav
i
ng
dim
ension
s
not
only
a
po
int
an
d
ave
r
ts
colli
sion
with
the
obsta
cl
es
us
ing
pr
e
ve
ntati
ve
m
easur
e.
Fi
g
ure
5
s
hows
eac
h
pa
rt
of
the
r
obot
co
ns
i
der
i
ng
i
n
a
vi
rtual
ci
rcle
avo
i
d
obsta
cl
e
an
d
th
e
op
e
n
po
i
nts
an
d
it
s
relat
ed
sub
goal
po
i
nts
is
sh
ow
n
in
Fig
ure
6
Ca
lc
ulati
ng
s
ob
go
al
po
i
nt
us
in
g
re
qu
i
re
d
m
et
ho
dolo
gy
is
ref
le
ct
ed
i
n
F
ig
ure
7
All
the
sub
goal
s
sit
ua
te
perpendicu
la
rly
rev
ers
e
di
recti
on
of
the
r
el
at
ed
op
e
n
points.
T
o
a
v
oi
d
the
ob
sta
cl
e
the
r
obot
has
to
tur
n
t
he
ra
kes
.
T
he
tur
ning
an
gle
and
dis
placem
ent
of
rob
ot’s p
os
it
ion
du
e
to
t
urn
is
shown i
n
Fi
g
ure
8.
Figure
5.
Whole r
obot insi
de
t
he virtual ci
rcle wit
h
safety
m
easur
em
ent
Figure
6.
O
bs
t
acl
es’
ope
n po
i
nt an
d su
b g
oal point
detect
ed by r
obot
3.1.
Al
go
ri
thm
study
and
analysis
Let
an
open
po
int i
s ide
ntifie
d by se
nsor at
a
ng
le
α.
As
per
[20]
β
i
= (α +β
i
-
1
)
%3
60
x
i+1,j
= x
+
d
k
cos
θ(x
s
)
(1)
y
i+1,j
= y+
d
k
sinθ(
y
s
)
(2)
Her
e
(
x,
y)
is
th
e
locat
io
n
of
se
ns
or
a
nd
(
x
i+1,j
,y
i+1,j
)
is
the
loc
at
ion
of
j
t
h
i
de
ntifie
d,
one
of
su
bse
que
nt
op
e
n
points
at
(
i+
1)
t
h
it
erati
on
a
nd
in
acc
ordan
ce
to wor
ld
s
co
ordinate
sy
stem
.
The
op
e
n
po
i
nt
ap
pea
rs
in
t
w
o
ways.
1.
I
f
ea
r
li
er
value
of
α
,
the
sen
sor
an
gle
α
-
1
sense
s
an
obst
ru
ct
io
n
an
d
2.
S
ub
se
quent
value
of
α
,
the
sens
or
an
gle
α
+1
senses
obstructio
n.
Eac
h
of
these
two
al
s
o
ha
ve
two
s
ub
cases
and
th
ose
are:
a)
>
9
0
0
an
d
b)
<=
90
0
wh
e
re
=
%
180
°
So
we
ca
n desc
ribe
as
foll
ows,
=
{
−
90°
,
Sx
=
1
,
Sy
=
1
if
>
90°
−
1
not
o
bst
a
c
le
−
90°
,
Sx
=
−
1
,
Sy
=
−
1
if
>
90°
−
1
is
obstacl
e
90°
−
,
Sx
=
1
,
Sy
=
−
1
if
≤
90°
−
1
not
obsta
cle
90°
−
,
Sx
=
−
1
,
Sy
=
1
if
≤
90°
−
1
is
obsta
cle
(3)
If
(
x
i+1,j
,y
i+1,j
)
is o
pe
n po
i
nt P t
hen
a
s
per
Fig
ure
7, c
oord
i
na
te
v
al
ue o
f
c
onnected
sub
goa
l po
i
nt Q is:
=
+
.
=
+
.
}
(4)
(Qx,Qy
)
will
be
the
ne
xt
centre
after
m
ov
e
of
the
virt
ual
ci
rcle
that
con
ta
ins
the
head
par
t
of
centi
pede
ins
pired
rob
ot.
I
n
a
ccorda
nce
t
o
F
ig
ure
4,
A
is
th
e
wh
e
el
axel
c
entre
a
nd
AC
i
s
the
dista
nce
a
m
on
g
the
virtu
al
ci
rc
le
centre
and
wh
eel
axel
ce
nt
re.
So,
the
val
ue
of
ne
xt
po
si
ti
on
of
w
heel
axel
centre
A
can
be
determ
ined
as,
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8938
IJ
-
AI
V
ol.
8
,
No.
2, June
201
9
:
95
–
106
10
2
=
+
.
=
+
.
}
(5)
Now
the
hea
d
par
t
of
rob
ot
w
il
l
co
m
pu
te
the
require
d
an
gle
of
ro
ta
te
to
m
ov
e
w
heel
axel
centre
a
nd
then
sta
rt
pro
gr
essi
ng
to
ne
xt
po
sit
io
n.
T
he
rem
ai
nin
g
par
t
of
the
robo
t
will
fo
ll
ow
it
pr
edecess
or
li
nk
s
coor
din
at
e
loc
at
ion
for
ne
xt
m
ov
e
as
wr
it
ten
in
a
lgorit
hm
.
Du
ri
ng
it
s
tur
ning
to
av
oid
obsta
cl
es
the
robo
t
get
s
disp
la
ce
d
from
it
s
d
m
in
li
ne,
wh
ic
h
it
has
to
cov
e
r
in
nea
r
fu
t
ur
e.
Ma
xim
um
it
turn
s
th
e
m
axi
m
u
m
it
t
akes
to
cov
e
r. Acc
ordi
ng Fig
ure
8,
if
Y
is t
he displa
ce
m
ent v
al
ue
then,
Y
=
(
F
+
R
)
sin
(
6)
=
sin
−
1
(
+
)
⁄
(7)
Her
e
(
F+R
)
is t
he
le
ngth
of ea
ch rake
and
is t
he
m
axi
m
um
p
er
m
issi
ble r
otati
on a
ng
le
of a r
a
ke
.
Irresp
ect
ive
of
m
ob
il
it
y
or
non
-
m
ob
il
it
y
ever
y
obsta
cl
e
is
sta
ti
on
ary
at
an
exact
point
of
tim
e.
I
f
the
po
sit
io
n
cha
ng
es
of
a
n
obsta
cl
e
then
it
is
m
ov
ing
ot
herwise
it
is
sta
t
i
c.
For
m
ov
ing
ob
sta
cl
e
the
po
s
sible
po
sit
io
n
in
f
utu
re
can
be
det
erm
ined
fr
om
i
ts
sp
eed
an
d
di
recti
on.
Fig
ure
9
il
lustrate
s
ho
w
to
a
vo
i
d
dynam
ic
ob
sta
cl
e
usi
ng
MC
SB
al
go
ri
thm
by
p
red
ic
ti
ng
the
virtu
al
ob
sta
cl
e
locat
ion
.
T
he
sta
te
d
schem
e
al
lows
the
rob
ot
to
av
oid
secur
el
y
sta
ti
c
ob
sta
cl
es
sho
w
n
in
Fi
gure
10
.
For
m
ob
il
e
obsta
cl
es
is
has
fi
nd
out
t
he
s
pee
d
of
that o
bs
ta
cl
e by
sen
sin
g
it
s
posit
ion at
tw
o
s
epar
at
e ti
m
e t
n+
1
an
d
t
n
w
he
re
t=
t
n+1
-
t
n
. Th
er
e
fore,
Figure
7.
O
pe
n p
oin
t t
o S
ub goal p
oin
t c
al
cu
la
ti
on
Figure
8.
Tu
r
nin
g an
gle a
nd d
i
sp
la
cem
ent d
ur
ing
t
urn
Figure
9.
A
voidin
g
a m
ov
in
g ob
sta
cl
e
by cal
culat
ing
virtu
al
obsta
cl
e twice
Figure
10.
A
voiding o
bs
ta
cl
e i
n
L
ocal Mi
nim
a
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
IS
S
N:
22
52
-
8938
Develo
pm
e
nt
of
p
at
h pla
nn
i
ng a
lg
or
it
hm
of c
entipede
ins
pir
ed
w
heele
d
r
ob
ot
...
(
Subir K
umar
Das
)
103
v
obs
=
(
O
t+1
−
O
t
)
t
(9)
wh
e
re
O
t+1
,O
t
are
O
bs
ta
cl
e
posit
ion
s
at
tim
e
instant
t
n+1
and
t
n
a
nd
v
obs
i
s
sp
ee
d
of
t
he
m
ob
il
e
ob
sta
cl
e.
By
so
lvi
ng
e
quat
ion
of
r
obot’s
directi
on
an
d
ob
sta
cl
e’
s
dir
e
ct
ion
the
pr
ob
able
posit
ion
of
colli
sion
(P
C
)
can
be
cal
culat
ed.
Th
e
nex
t
ta
sk
is
to
fin
d
out
w
he
ther
ther
e
will
be
a
real
colli
sion
or
not
at
P
C
.
If
there
will
be
a
colli
sion
the
n
t
he
posit
io
n
an
d
it
s
su
r
r
ounding
will
be
ide
nt
ifie
d
as
virt
ua
l
ci
rcle.
If
P
R
(rx
n+1
,ry
n+
1
)
be
c
urren
t
locat
ion
of the
rob
ot at t
n+1
. T
hen,
D
C
=
√
(
Cx
−
rx
n+1
)
2
+
(
Cy
−
ry
n+1
)
2
(10)
w
he
re
(Cx,C
y)
is the c
oor
din
a
te
p
osi
ti
on
of P
C
. By t
he value
of D
C
r
obot th
en dete
rm
ines,
R
TC1
=
D
C
v
⁄
(11)
R
TC2
=
(
n
∗
2r
+
D
C
)
v
⁄
(12)
O
TC
=
D
C
v
obs
⁄
(13)
The
velocit
y
of
r
obot
is
v.
R
T
C1
and
R
TC2
are
tim
e
of
enteri
ng
at
P
C
an
d
e
xiti
ng
from
P
C
resp
ect
ively
and
O
TC
is
th
e
t
i
m
e
the
ob
st
acl
e
will
ta
ke
to
ar
rive
at
c
ol
li
sion
poi
nt.
N
is
num
ber
of
li
nk
s
or
ra
kes
of
t
he
rob
ot
and
r
is
the
rad
i
us
of
vi
rtual
ci
rcle.
O
TC
li
es
tim
e
ran
ge
R
TC1
and
R
TC2
m
eans
they
will
be
at
P
C
i
n
the
sam
e
tim
e
per
iod
i
n
nea
r
f
uture.
S
o
the
po
i
nt
will
be
ind
ic
a
te
d
as
virt
ual
obsta
cl
e
an
d
dis
torted
tra
j
ect
ory
will
be
s
hap
e
d
with
safety
m
easure
.
H
ow
t
he
r
ob
ot
will
avo
i
d
l
ocal
m
ini
m
a
pr
oble
m
is
il
lust
rated
in
Fig
ure
9.
If
the
rob
ot
fail
s
to
fin
d
an
op
e
n
point
in
dire
ct
ion
F
OV
the
head
will
ro
ta
te
righ
t
a
nd
le
ft
as
m
uch
as
po
ssible,
scan
the
area
t
ake
op
e
n
po
i
nts.
I
f
it
sti
ll
fails
then
it
has
to
ro
ta
te
the
w
hole
robo
t.
The
F
ig
ure
11
s
how
s
ho
w
to
av
oid
a
n
obsta
cl
e
with
saf
et
y
m
easur
em
e
nt.
He
re
in
this
fig
ur
e
it
is
de
s
cribe
d
f
or
a
sin
gle
li
nk.
But
f
or
any
lin
ks
it
is a
pp
li
cable as al
l t
he
li
nk
s
foll
ow th
e sam
e p
at
h
to
reach
d
est
in
at
ion.
Figure
11.
O
bst
acl
e avo
i
dan
c
e w
it
h safet
y m
easur
em
ent
3.2.
To
ta
l
tim
e a
n
d pat
h len
gt
h
c
alculat
i
on
If
v
i
an
d
i
are
the
s
peed
and
a
ngular
velocit
y
at
it
h
it
erati
on
the
n,
ti
m
e
ta
ken
in
m
ov
ing
and r
otati
ng [2
0]:
=
∑
⁄
=
0
=
∑
⁄
=
0
So
t
he
c
os
t f
un
ct
ion
in
term
s o
f
tim
e fo
r
one
li
nk
s
is:
=
∑
(
⁄
+
⁄
)
=
0
+
∑
=
0
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8938
IJ
-
AI
V
ol.
8
,
No.
2, June
201
9
:
95
–
106
104
w
he
re
DOT is
the tim
e taken t
o
ide
ntify m
ob
il
e obstac
le
a
n
d i
ts directi
on of
progress
and s
pee
d
.
So
t
he
total
c
ost
in
te
rm
s o
f
ti
m
e fo
r
all
the l
ink
s
or the
wh
ole ro
bo
t i
s:
=
∑
1
3.
3
.
Simul
at
io
n
resul
ts
How
a
rob
ot
is
us
in
g
M
od
i
fiedCrit
ic
al
-
Sn
a
kebu
g
al
gorith
m
to
sta
y
away
fr
om
sta
ti
onary
an
d
non
-
sta
ti
on
ary
obst
acl
e
sh
own
i
n
the
Fig
ure
12.
Ther
e
a
re
twel
ve
sepa
rate
sn
a
p
shots
of
twel
ve
sepa
rate
m
ome
nts
in
sa
m
e
env
ir
on
m
ent.
The
gr
ay
colo
ured
,
rectangula
r
s
hap
e
d
ob
j
ect
is
non
-
sta
ti
onar
y
ob
sta
cl
e
and
sta
tic
ob
sta
cl
es
are
of
gr
e
en
re
ct
an
gula
r.
T
he
al
go
r
it
h
m
is
m
od
el
l
ed
us
i
ng
Pyt
ho
n
3.5
on
windows
7
platfo
rm
.
In
te
l
cor
e
i
3
-
2350@
2.30
G
hz
Lapt
op
with
2
GB
RAM
is
us
e
d.
The
re
d
ci
rcle
at
the
top
a
nd
foot
are
the
go
al
and
init
ia
l po
ints
re
sp
ect
ively
. T
he
p
at
h r
obot tr
av
el
le
d
is sh
own
us
in
g
Bl
ack
li
ne
.
Fi
gure
12.
Sn
a
ps
hots
of tra
j
ec
tory
gen
e
rated
us
in
g
m
od
ifie
dc
riti
cal
-
sn
ake
bug al
gorithm
on 2D
p
la
ne
4.
CONCL
US
I
O
NS
A
ND FUT
UR
E
WO
RK
The
earli
er
w
orks
is
f
ur
th
er
c
on
ti
nue
d
in
this
pap
e
r.
T
he
f
or
em
os
t
intenti
on
is
to
buil
d
a
local
path
plan
ning
al
gor
it
h
m
on
sim
pl
e
sens
or
base
d
f
or
bi
o
-
in
sp
ir
ed
ce
ntipe
de
li
ke
m
ulti
rak
e
rob
ot.
T
he
al
gorithm
m
akes
an
ef
fort
to
t
hink
a
bout
the
restrict
ion
s
of
rob
ot’s
dim
ension
.
T
o
c
om
plete
the
ta
sk
us
in
g
t
his
bu
g
al
gorithm
the
rob
ot
re
qu
ire
s
ver
y
li
tt
le
pr
ior
i
nfor
m
at
i
on.
Me
asu
rin
g
t
he
safety
pr
ecauti
on
m
akes
th
e
al
gorithm
m
or
e
eff
ect
ive
.
T
hi
s
al
go
rit
hm
helps
to
a
vo
i
d
a
ver
y
sm
al
l
th
e
gap
or
to
pa
ss
thr
ough
sm
al
l
gap
wh
ic
h
is bi
g
e
noug
h
t
o
pass.
I
n
case
le
ss
c
om
plex
env
ir
on
m
ent
or
e
nv
ir
onm
ent
with f
e
w
nu
m
ber
s
of o
bs
ta
cl
e,
Evaluation Warning : The document was created with Spire.PDF for Python.