TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.6, Jun
e
201
4, pp. 4596 ~ 4
6
0
2
DOI: 10.115
9
1
/telkomni
ka.
v
12i6.544
2
4596
Re
cei
v
ed
De
cem
ber 2
8
, 2013; Re
vi
sed
F
ebruary 27,
2014; Accept
ed March 1
4
, 2014
A Novel Control Architecture for Missio
n
Re-Planning
of AUV
Rubo Zh
ang
1,2
, Haibo Tong
1*
, Changti
ng Shi
1
1
Colle
ge of Co
mputer Scie
nc
e and T
e
chno
l
o
g
y
, Harb
in En
gin
eeri
ng U
n
iv
ersit
y
,
Harbi
n
, 150
00
1, Chin
a
2
Colle
ge of Ele
c
tromecha
nica
l
&
Information E
ngi
neer
in
g, Dal
i
an N
a
tion
aliti
e
s Universit
y
,
Dali
an, 11
66
00
, China
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: tongha
ib
o08
01@
hotmai
l
.co
m
A
b
st
r
a
ct
A hi
erarch
ical
contro
l
archit
ecture for
mis
s
io
n
re-p
la
nni
n
g
of
auto
n
o
m
ous
un
derw
a
te
r veh
i
cl
e
(AUV) navig
ati
ng in u
n
certai
n oc
ea
n envir
on
me
nt is pre
s
ented i
n
this pap
er. T
he propos
ed co
mp
o
nent-
orie
nted co
ntro
l architectur
e
structur
ed is
ma
de of three p
a
r
t
s: situati
on re
ason
ing, re-
p
la
nni
ng trig
ger a
n
d
hier
archic
al re-
p
la
nni
ng l
a
yer. Situatio
n reas
o
n
in
g us
in
g the
unstructure
d re
al-w
ord infor
m
ation o
b
tai
ned
by
sorts
of
sensor
detectes and recog
n
i
z
e
s
un
certain ev
e
n
t.The re-
p
la
nn
ing
trigger
deci
d
e
s
the re-p
lan
n
i
n
g
level
by th
e ev
ent types
an
d i
n
flue
nce
de
gre
e
. Hier
a
rchic
a
l
re-pla
nn
ing
lay
e
r conta
i
ns
mis
s
ion r
e
-pl
a
n
n
in
g
,
task re-pl
ann
in
g an
d b
ehav
ior
re-pl
ann
in
g. D
i
fferent re-p
la
n
n
in
g lev
e
l
dep
e
nds o
n
the r
e
s
u
lt of re-p
la
nni
ng
trigger. Prel
iminary versi
ons
of the architecture
hav
e bee
n inte
grate
d
and teste
d
in a simulati
o
n
envir
on
me
nt. Experi
m
ent in
d
i
cates t
hat the
nove
l
contro
l a
r
chitecture
c
a
n
imple
m
ent
mi
ssion re-
p
l
ann
i
n
g
steady an
d safty.
Ke
y
w
ords
: aut
ono
mous u
nde
rw
ater vehicle,
cont
rol arc
h
itec
ture, missi
on r
e
-pl
ann
in
g
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Develo
pment
s in A
u
tonom
ous Unde
rwa
t
er Vehi
cle
(A
UV)
have b
e
en of g
r
e
a
t in
terest to
many re
sea
r
che
r
s, e
ngin
eers and
scientist [1-3
]. The capabiliti
e
s of AUVs
as well as t
heir
missi
on
req
u
i
reme
nts h
a
ve bee
n in
creased. Re
ce
nt advan
ce
s in auton
om
ous
und
erwa
ter
vehicle te
chn
o
logy have l
ed to their
use in
a n
u
m
ber of milit
ary and
civilian appli
c
atio
ns
inclu
d
ing
a
n
ti-su
b
ma
rine
wa
rfare, oil
fi
eld
su
rveys,
oceano
gr
a
phic re
se
arch
or
maintena
nce/monitori
ng of unde
rwater
structu
r
e
s
amo
ng ot
hers un
d
e
rwater
scen
ario
s
Re
sea
r
ch on
autonomo
u
s underwate
r vehicle
o
w
n the com
m
on
control probl
em with
other air, lan
d
and se
a surface unma
nned vehi
cle
becau
se of
the dynamic and un
ce
rtain
environ
ment.
But in
marin
e
envi
r
onm
ent, b
e
sid
e
s
re
qu
iring
high
-di
m
ensi
onal
and
comp
utationa
lly intensive
sen
s
o
r
y data
for
real
-wor
l
d
missio
n ex
ecutio
n, sta
b
i
lity of sona
r
and
rand
om o
c
curren
ce is m
a
ke it more difficult
to develo
p
control arch
itecture fo
r AUV.
We
present
a hyb
r
id, hie
r
archical a
r
chi
t
ec
ture
for mi
ssi
on
re
-pla
n
n
ing
of auto
n
o
mou
s
unde
rwater v
ehicl
e. Ou
r g
oal is to
dev
elop n
o
vel
co
ntrol a
r
chitect
u
re to
reali
z
e
the missio
n
re-
planni
ng
wh
en the
p
r
evi
ous mi
ssi
on
plan
cann
ot
execute
co
rrectly. T
h
ro
u
gh the
situat
ion
rea
s
oni
ng pe
rceive ab
norm
a
l events an
d
the re-pl
anni
ng trigge
r de
cides the re-pl
annin
g
level
2. An Ov
er
vi
e
w
of Control Architecture
A co
ntrol
architecture [4]
is the
p
a
rt
o
f
the robot
control
sy
ste
m
which m
a
ke
s the
deci
s
io
ns. Th
e first attemp
t at building contro
l archit
ecture for aut
onomo
u
s u
n
derwate
r vehi
cle
bega
n aro
u
n
d
1990
s. Tra
d
itional archit
ecture relie
d on a cent
rali
zed
worl
d model for verif
y
ing
sen
s
o
r
y info
rmation a
nd
g
enerating
acti
ons in th
e world
mod
e
l, followin
g
the
sen
s
e,
plan,
and
act patte
r. Th
e de
sign
of the cl
assi
cal
control
a
r
chitecture
wa
s b
a
se
d on
a to
p-do
wn
struct
ure
[5]. The
seq
u
ence of
pha
se in t
r
aditio
n
a
l deli
berativ
e control a
r
chitecture i
s
shown in
Fig
u
r
e
1.
The
comm
on
pro
b
lem
s
for this a
r
chitecture a
r
e th
at the integ
r
atio
n re
pre
s
e
n
tation of the
re
al
worl
d is extre
m
ely difficult and the
sen
s
or data
ca
n o
n
ly use
s
du
ri
ng the world
model a
nd n
o
t
durin
g the pla
n
executio
n. It is dange
rou
s
in dynami
c
marin
e
enviro
n
ment.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
De
sign a
nd Im
plem
entation of Universal
Serial Bus T
r
anscei
v
e
r
wit
h
Verilog (Liq
un Xu)
4597
Figure 1. Phase of Tra
d
itio
nal Co
ntrol Archite
c
ture
The b
ehavio
r-ba
sed
a
r
chitecture u
s
e
d
a set of p
a
ral
l
el beh
aviors
whi
c
h
rea
c
te
d to the
worl
d environ
ment su
gge
st
ing the
re
sp
o
n
se the
rob
o
t shoul
d take t
o
finish the b
ehavior
(see
as
Figure 2
)
. Th
e be
havior-b
a
se
d a
r
chitecture i
s
fa
st a
nd
rea
c
tive a
nd
solve
s
th
e p
r
oble
m
wi
th
worl
d mo
deli
ng o
r
real ti
me p
r
o
c
ess.
Ho
wever wh
en trying
to
carry o
u
t lon
g
-rang
e mi
ssions,
there a
r
e so much limitatio
ns an
d it is
difficult to optimize the robot b
ehaviors.
Figure 2. Behavior-ba
se
Control Archite
c
ture
Most of today
’s archite
c
ture for autono
mous
rob
o
tics is hyb
r
id an
d stru
ctured i
n
three
layers: the re
active layer, the co
ntrol ex
ecutio
n
layer,
and the delib
erative layer (see a
s
Fig. 3).
It integrate the advantag
es of previous t
w
o, but
it is complex to ha
ndle dynami
c
and un
certai
n
environ
ment
and missio
n re-pla
nnin
g
.
Figure 3. The
Hybrid Control Archite
c
ture
3. Our Propo
sal
In ord
e
r to
solve ab
ove p
r
oble
m
s men
t
ioned i
n
p
r
e
v
ious
se
ction
.
We p
r
op
ose
a n
o
val
control archit
ecture for mi
ssi
on pla
nnin
g
of aut
onom
ous u
nde
rwat
er vehi
cle. It is a hybird a
nd
hiera
r
chi
c
al framework (se
e
as Fi
gure 4
)
. The p
r
op
osal framwork
contain three l
a
yers:
situati
o
n
rea
s
oni
ng lay
e
r, re-pl
anni
n
g
trigg
e
r
and
hiera
r
chi
c
al
re-pla
nnin
g
la
yer. Situation
rea
s
o
n
ing l
a
yer
usin
g the u
n
s
tru
c
tured
re
al-word information
o
b
tai
ned by
sen
s
ors dete
c
te
s and
re
cog
n
i
z
e
s
uncertain
ev
ent. Accordi
ng the
even
t types a
n
d
influen
ce d
egre
e
, the
re-pla
nnin
g
tri
gger
deci
d
e
s
the re-pla
nnin
g
le
vel. Hiera
r
chi
c
al re-pl
anni
n
g
layer contai
ns mi
ssi
on re
-plan
n
ing, ta
sk
re-plan
n
ing
a
nd b
ehavio
r
re-plan
n
ing.
Acco
rdi
ng th
e
re-plan
n
in
g level
gen
e
r
ated
from
re-
planni
ng trigg
e
r, hierarchi
c
al re-plan
n
ing
laye
r will sel
e
ct co
rrespon
ding re
-pl
anni
ng layer.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4596 – 4
602
4598
Figure 4. Our Propo
sal Co
ntrol Architect
u
re for Mi
ssio
n Re-plan
n
in
3.1. Situatio
n Reas
oning
Situation Rea
s
oni
ng i
s
o
n
e
of the mo
st
core
part i
n
th
e propo
sal
co
ntrol a
r
chitect
u
re [6].
This layer ge
nerate
s
the i
n
fluen
ce
deg
ree
of ev
e
n
ts acco
rdi
ng
u
s
er in
put, en
vironme
n
t se
nso
r
data a
nd gl
o
bal
kno
w
led
g
e
such a
s
ta
sk ty
pe exe
c
uting, data
b
a
s
e
of plan
m
e
thod
and A
U
V
perfo
rman
ce
whi
c
h u
s
u
a
lly never
ch
ang
ed. The
re
su
lt of this layer, influence de
gree
of event
s,
is se
nt to the re-plan
n
ing tri
gger.
Situation Mo
del receives
the un
ce
rtain
info
rmatio
n from
ta
sk executio
n
info
rmation,
internal
state
data of AUV
and e
n
viron
m
ent se
ns
or d
a
t
a to modele
d
uncertain
events a
c
cordin
g
to their prio
rity and their n
a
ture. Ta
sk e
v
ent det
ectio
n
detect
s
the
implementati
on and p
r
og
ress
of the AUV task. Enviro
nm
ent event det
ection p
e
rc
e
p
t
s the enviro
n
m
ent relative
to the assign
ed
missi
on, i
n
cl
uding
the
en
vironme
n
tal
status,
a
ttrib
utes, a
n
d
dy
namics. Stat
e event
dete
c
tion
delete faults
of AUV various sen
s
ors.
Dete
cte pa
rt inclu
d
ing Ta
sk dete
c
tion,
Envi
ronme
n
t detectio
n
an
d State detection, is
use
d
to
han
dle with
the
un
certai
n eve
n
t
informatio
n a
nd b
e
a
b
le to
pre
d
ict th
e in
fluence d
egre
e
of event for completion of the task.
Combi
ne
with glob
al kno
w
led
ge, the i
nput
of un
ce
rtain event se
e as
Table
1
and lthe
output of situ
ation re
asoni
ng contain
s
e
v
ent ty
pe, probability an
d
Influence de
g
r
ee o
n
the ta
sk,
see a
s
Tabl
e 2:
Table 1, Input
of Unce
rtain
Event
Task information
Status information of AUV
Environment
T
a
sk ty
pe
Position of target
Planning informa
t
ion
Cabin leaks
Batter
y
comp
art
m
ent
leaks
Fault of lo
w
voltage
Lo
w
voltage is low
Fault of high voltage
Fault of depth
ga
uge
Etc...
Status informat
ion of AUV
Obstacle (statics or dy
nam
ics,
mor
e
or
less)
target inform
ation
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
De
sign a
nd Im
plem
entation of Universal
Serial Bus T
r
anscei
v
e
r
wit
h
Verilog (Liq
un Xu)
4599
Table 2. Outp
ut of Situation Rea
s
oni
ng
Event t
y
pe
Event name
Event
Probabilit
y
Influence degree
on
task
Ocean
environment
Random object
0~1
0~1
Target unachieved
0~1
0~1
Off Rout
e
0~1
0~1
Target Lost
0~1
0~1
AUV status
Energ
y
sho
r
tage
0~1
0~1
GPS correction
0~1
0 or1
Fault of pro
peller
0~1
0~1
Uncer
t
ain task
Task paramete
r
s changed
0~1
0 or 1
Task t
y
pes chan
ged
0~1
0 or 1
Re-plan
n
ing t
r
igge
r receives the
re
sult
of
situation
reasonin
g
. When the follo
wing fo
ur
sort
s event (not limited) h
appe
ned, the
re-pl
anni
ng
trigge
r will b
e
trigged. First
one is that the
planni
ng mo
nitoring o
b
se
rve the plan
prog
re
ss ha
s
large d
e
viation with the o
r
iginal pl
an; the
se
con
d
on
e i
s
that the
external
en
ciron
mment o
r
task target ha
s
cha
nge
d; the
third o
ne i
s
t
hat
the internal
status of AUV chan
ge a lot
,
the
task ca
n impleme
n
t anymore; the last one is a
n
y
unpredi
ctable
events hap
p
ened.
3.2. Hierarch
ical Re-pla
n
n
ing La
y
e
r
Missio
n
re
-pl
annin
g
i
s
im
plemente
d
m
a
inly
be
ca
use the
missio
n target
has
cha
nge
d
whi
c
h may
caused by op
erato
r
s th
rou
gh the u
s
e
r
interface al
so be
cau
s
e t
hat the plan
ning
monitori
ng m
odule fo
und t
he statu
s
of
AUV ha
s diffe
rent
with the
origin
al an
d
then dete
r
mi
ne if
missi
on
re
-pl
annin
g
shoul
d be im
plem
ented. Be
si
d
e
s
a
bove, wh
en
the enviro
n
ment whi
c
h the
planni
ng relie
s on
ha
s be
e
n
ch
ang
ed,
missi
on
re-pl
annin
g
will al
so b
e
exe
c
ut
ed. Ne
w mi
ssion
planni
ng ca
n
improve AUV’
s efficien
cy.
Task re-plan
n
ing i
s
carrie
d out mainly
becau
se the
missi
on
re-pl
annin
g
. But missi
on
re-
planni
ng
will
spe
n
t mu
ch
times and
e
n
e
rgy. In
ord
e
r
to i
m
prove
system
flexibi
lity and
rea
c
ti
on
spe
ed, task re-pla
nnin
g
lo
wer tha
n
mission re-plan
n
ing is al
so ne
eded. When t
he statu
s
of AUV
or environme
n
t chan
ge
s a little, task re-planni
ng is e
noug
h.
Behavior
re-p
lannin
g
is the
lowe
st level for re
-pla
nnin
g
. When it re
ceive the
co
mmand
from mi
ssi
on
re-plan
n
ing
a
nd task
re-pl
annin
g
. It will adju
s
t the b
ehavior
or
action se
que
nce to
finish
ne
w ta
sk. A
nothe
r
situation i
s
tha
t
whe
n
im
ple
m
enting
the
origin
al a
c
tio
n
with
no
cha
nge
can
not finish
the sub
-
task,
the behavio
r re-pla
nnin
g
wil
l
also sta
r
t
4. Experiment
The exp
e
rim
ent is u
s
ed
to dem
on
strate
s the
a
d
vantage
fro
m
propo
se
d
co
ntrol
architectu
re. Experiment
s inclu
de com
m
on navigati
on, navig
aito
n in
cu
rrent, static ob
sta
c
le
avoidan
ce a
n
d
target un
re
ach. A co
ntra
st experim
ent
is also impl
e
m
ented.
Navigatio
n in
uncertain
u
nderwate
r e
n
v
ironme
n
t is familiar. Th
e Figu
re 5
show th
e
AUV’s navig
a
t
ion without u
n
ce
rtain even
ts.
Figure 5. AUV’s Navigatio
n in Und
e
rwa
t
er Environm
ent
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4596 – 4
602
4600
Whe
n
the
r
e
a
r
e
ob
stacl
e
s.
The
co
ntrol
a
r
chite
c
tu
re
wil
l
sta
r
t the
mission
re-pl
anni
ng a
n
d
carry out the
obsta
cle avoi
dan
ce, at the same
time
s
the AUV deviates from the
predet
ermin
e
d
route an
d fart
her a
w
ay, se
e as Figu
re 6.
Figure 6. AUV Obsta
c
le Avoidan
ce
A cont
ra
st na
vigation expe
riment i
s
al
os
pro
p
o
s
ed
in
Figure 7. T
h
e
left one i
s
used th
e
control archit
ecture presen
ted in the pap
er for mi
ssio
n
re-pl
anni
ng.
Figure 7. Con
t
rast Navigation Experime
n
t
Navigation in current is
shown
in Figure 8. The control ar
chitecture will carry out th
e
missi
on re-pl
annin
g
.
Figure 8. Nav
i
gation in Current
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
De
sign a
nd Im
plem
entation of Universal
Serial Bus T
r
anscei
v
e
r
wit
h
Verilog (Liq
un Xu)
4601
Ran
dom
ob
stacle
s a
ppe
ared event i
ndi
cate
s t
he
dist
ances f
r
om
o
b
sta
c
le
s is le
ss than
s
a
fe
dis
t
anc
e
, whic
h is
threat toAUV.
AUV is more
cl
oser to
obstac
les, the hi
gher the probability
of rand
om o
b
s
tacl
es
app
e
a
red
event h
appe
ned i
s
.
The
control a
r
chite
c
tu
re d
e
tecte
s
the e
v
ent
and trigg
e
ts
missi
on re-pl
annin
g
, see a
s
Figu
re 9.
Figure 9. Ran
dom Ob
st
a
c
les Appe
ared
Event
Key point un
rea
c
he
d eve
n
t mean
s tha
t
AUV can n
o
t rea
c
h the
target point
whi
c
h is
covered
by o
b
sta
c
le
s, or i
n
valid pla
nni
ng a
c
ti
on
s. It will ge
nerate
uncertain
eve
n
t and tou
c
h
off
the mission re-pla
nnin
g
, and re
-pla
n an
other pla
n
to achi
eve other target, see a
s
Figu
re 10.
Figure 10. Key Point Unrea
c
he
d Event
5. Conclusio
n
In the pa
per, we p
r
e
s
ent
novel
control ar
chite
c
ture for
re-plan
n
ing of
auto
nomo
u
s
unde
rwater v
ehicl
e. Prelim
inary version
s
of
the archi
t
ecture
have
been inte
grated and
analy
s
is
in a marin
e
simulation
environ
ment.
The re
su
lt
demon
strates the benefits of the co
ntrol
architectu
re
with re
-pla
nni
ng feature. F
u
ture work wi
ll focus o
n
the real ma
rine
environm
ent
to
verify the pra
c
ticality and e
fficiency of this pro
p
o
s
ed
control a
r
chite
c
ture.
Ackn
o
w
l
e
dg
ements
This work was finan
cially
suppo
rted b
y
the National Natural Scien
c
e Fo
un
dation of
Chin
a (No.60
9750
71) a
nd
(No.6
110
000
5).
Referen
ces
[1]
Muhamm
ad N
a
siru
ddi
n Mah
y
u
d
d
i
n, Moh
d
Rizal Ars
h
a
d
.
Classes
of Co
ntrol Archit
ectu
res for AUV: A
Brief Survey.
I
n
ternati
o
n
a
l C
onfere
n
ce
on
Und
e
r
w
ater S
ystem T
e
chnolo
g
y
: T
heor
y
an
d App
licati
o
n
s
(USYS08). Bal
i
.
2008.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4596 – 4
602
4602
[2]
DP Eickstedt, Scott R.
The
backse
at control arch
itecture
for
autono
mo
us robotic ve
hi
cles: A cas
e
study w
i
th the Iver2 AUV.
OCEANS 2009,
M
T
S/IEEE Bil
o
xi - Mari
ne Technology
for Our Future:
Globa
l an
d Loc
al Ch
all
eng
es. Bilo
xi. 2
009; 1-
8.
[3]
Lei Z
h
an
g, Da
-pen
g Ji
ang, J
i
n-xin Z
h
ao. T
he Ba
sic
Co
ntrol S
y
stem
of a
n
Ocea
n E
x
pl
oratio
n AUV
.
Appl
ied Mec
h
a
n
ics an
d Materi
als.
201
3; 411-
414: 17
57-
176
1
[4] R
Arkin.
Editor
s.
Behavior-B
a
s
ed Ro
botics.
Boston
MA: MIT
Press. 1998.
[5]
Narcis P
a
lom
e
ras, Andr
es El-Fakdi, Marc
Carrera
s, P
e
re
Rida
o
. COLA
2
:
A Contro
l Ar
chitecture
for
AUVs.
IEEE Journa
l Of Oceanic Engi
neer
in
g.
2012: 3
7
(4): 6
9
+
5
-71
6
.
[6]
Lili
Yi
n , Rubo
Zhang, Hen
g
w
e
n
Gu.
Situ
ation re
ason
in
g for an adj
ustabl
e auto
n
om
y
s
y
stem.
Intelli
gent C
o
mputin
g an
d Cyb
e
rnetics.
20
12;
5(2): 226-2
38.
Evaluation Warning : The document was created with Spire.PDF for Python.