Intern
ati
o
n
a
l Jo
urn
a
l
o
f
R
o
botics
a
nd Au
tom
a
tion
(I
JR
A)
Vol
.
3
,
No
. 2,
J
une
2
0
1
4
,
pp
. 13
1~
13
8
I
S
SN
: 208
9-4
8
5
6
1
31
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
/
IJRA
Multisen
sor Dat
a
Fusion and Integrati
o
n f
o
r M
o
bile
Robots: A Review
KS Na
gl
a
*
,
M
o
i
n
Uddi
n
*
*, and
Di
l
b
ag
Si
ngh
*
*Facult
y
of Instr
u
m
e
ntation
and
Control
Engine
e
r
ing, Dr
BR Am
bedkar N
a
tion
a
l
Institute
of Technolog
y
J
a
landh
ar-India
**Facult
y
of
Ele
c
tri
cal
Eng
i
neer
i
ng, Delh
i
Tec
h
n
o
logic
a
l Univ
ersit
y
De
lhi-Indi
a
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Aug 4, 2013
Rev
i
sed
Ap
r 2, 20
14
Accepted Apr 18, 2014
One of the most important and us
eful
feature of
autonomous mobi
le robots is
their ab
ility
to adopt themselves to
operate in
unstructured environment.
Today
robo
ts are performing autonomously
in
industrial oor
, o
ce
environments, as well as in cro
w
ded pub
lic p
l
aces.Th
e
basic requirement o
f
an intellig
ent
m
obile robot is
to de
velop
an
d m
a
intain lo
calization and
mapping parameters to
complete the
co
mplex
missions. In such situations
,
s
e
veral
dicu
lti
es
aris
e
due
to th
e in
accur
a
c
i
es
a
nd uncer
tain
ties
in s
e
ns
o
r
m
eas
urem
ents
. Various
techniq
u
es
are ther
e to
handle s
u
ch nois
e
s
where the
m
u
ltisensor dat
a
fusion is no
t
the
exc
e
ption
a
l one
. During
t
h
e last
two
decad
es, m
u
ltisensor data fusions in
m
obile robots becom
e
a dom
inant
paradigm due to its potential
adva
ntages lik
e reduction in
uncertain
ty
,
increase in accu
racy
,
and redu
ction of
cost.This paper presents
the detail
review of m
u
ltisenosr data fusion a
nd its applications for autonom
ous
m
obile.
Keyword:
Aut
o
n
o
m
ous M
obi
l
e
R
o
b
o
t
s
Mu
ltisen
sor Data
Fu
si
o
n
Mu
ltisen
sor In
t
e
g
r
ation
Copyright ©
201
4 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
:
KS Na
gla
A
liatio
n
:
Dr B
R
A
m
b
e
d
k
a
r
Natio
n
a
l In
stitu
t
e
of Tech
no
logy, Jaland
h
a
r-Ind
i
a
Ad
dr
ess:
De
par
t
m
e
nt
of
In
st
ru
m
e
nt
at
i
on a
n
d
C
ont
r
o
l
E
n
gi
ne
eri
n
g, R
o
b
o
t
i
c
s
Lab
.
Dr B
R
A
m
bedkar
Nat
i
o
nal
In
stitu
te o
f
Tech
no
log
y
, Jaland
h
a
r-In
d
i
a
Pho
n
e
:+9
1
-1
81
-27
903
02
Em
ail: n
a
g
l
ak
s@nitj
.ac.in
1.
INTRODUCTION
Ro
bo
tics is an
ex
trem
ely challenging resea
r
ch ar
ea which
deals with va
rious iss
u
es like
structural
d
e
sign
, m
o
b
ilit
y, con
t
ro
l, lo
calizatio
n
and
map
p
i
ng
, etc.
In th
e last t
w
o d
ecad
e
s, sev
e
ral n
e
w tech
nolo
g
i
es
have
bee
n
e
xpl
ore
d
t
o
i
m
pro
v
e
t
h
e ab
o
v
e i
s
s
u
es. T
o
day
r
o
b
o
t
s
are a
b
l
e
t
o
navi
gat
e
aut
o
n
o
m
ousl
y
i
n
di
f
f
ere
n
t
envi
ro
nm
ent
s
such as
dy
nam
i
c or st
at
i
c
, i
n
d
o
o
r
o
r
o
u
t
d
oor, etc. But still there ar
e se
ve
ra
l
open c
h
al
l
e
n
g
e
s t
h
at
n
eed to
b
e
con
s
id
ered fo
r
fu
rt
h
e
r
d
e
v
e
lopmen
ts. Mu
ltis
en
sor
d
a
ta sen
s
o
r
fusion
techn
i
qu
e is an
essen
tial
p
r
o
cess to
im
p
r
ov
e th
e au
tono
m
o
u
s
cap
ab
ilities o
f
th
e m
o
dern
robo
ts. Th
ere is a con
s
id
erab
le con
t
ribu
tio
n
i
n
this resea
r
ch area that shows
how m
easurem
e
n
ts from
di
ffe
r
e
nt
sen
s
o
r
s ca
n
be c
o
m
b
i
n
ed t
oget
h
er t
o
m
a
ke t
h
e
syste
m
m
o
re reliab
l
e and
accu
rate.In th
e view
o
f
th
is,
the literatu
re surv
ey i
n
t
h
is
pap
e
r is
d
i
v
i
d
e
d
i
n
t
o
d
i
fferen
t
sectio
n
s
. Th
e in
itial sectio
n d
e
als with
an overv
iew
o
f
au
t
o
no
m
o
u
s
m
o
bile rob
o
t
s and ro
le
of
m
u
l
tisen
o
s
r
d
a
ta fu
sion
. In
this sectio
n
,
m
u
lt
isen
o
s
r d
a
t
a
f
u
si
on an
d i
n
t
e
gr
at
i
on i
s
di
ffe
re
nt
i
a
t
e
d and
rev
i
ewed
in
d
e
tail. secon
d
p
a
rt
o
f
t
h
e p
a
p
e
r deals wi
th
th
e literatu
re sh
owing
v
a
ri
o
u
s
adv
a
n
t
ag
es o
f
m
u
ltisen
osr d
a
ta
fusi
on
i
n
m
obi
l
e
r
o
b
o
t
s
.
The
l
a
st
sect
i
on
o
f
t
h
e
pape
r e
x
pl
o
d
es
vari
ou
s se
n
s
or
f
u
si
o
n
al
go
ri
t
h
m
s
.
2.
AUTONOMOUS MOBILE ROBOTS
AN OVERVIEW
To
day
R
o
bot
i
c
t
echn
o
l
o
gy
ha
s m
oved f
r
o
m
t
h
e i
n
dust
r
i
a
l
m
a
nufact
uri
ng
pl
ant
s
t
o
t
h
e
u
n
p
r
e
d
i
c
t
a
bl
e
com
p
l
e
x envi
r
onm
ent
.
Due t
o
hi
g
h
dem
a
nd of se
rvi
ce r
o
b
o
t
s
, t
h
e t
r
a
d
i
t
i
onal
i
ndu
st
ri
al
rob
o
t
s
are bei
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
089
-48
56
IJR
A
V
o
l
.
3,
No
. 2,
J
u
ne 2
0
14:
1
3
1
– 13
8
13
2
rep
l
aced
b
y
the em
erg
i
n
g
auto
no
m
o
u
s
in
tel
lig
en
t m
o
b
ile ro
bo
ts. Su
ch intellig
en
t robo
ts h
a
v
e
t
h
e ab
ility to
adjust their
be
havi
or a
u
tonom
ously
, according to the environm
ent. High
degree of a
u
tonom
y is de
sired i
n
vari
ous m
obile
robot applications s
u
c
h
as: s
p
ace expl
or
ation, fl
oors clea
ning, m
o
wing
lawns
,
and m
a
terial
trans
p
ortation, etc.
In
these
applications, the
workplaces a
r
e
high
ly challenging a
n
d often contain
untidy
a
nd
u
npr
ed
icted
p
hysical en
v
i
r
onmen
t. I
n
su
ch
u
npr
ed
icte
d e
n
vi
r
onm
ent
,
t
h
e necessary
processes that
m
u
st be
coo
r
di
nat
e
d
t
o
pe
rf
orm
t
h
e
desi
re
d t
a
s
k
s
are se
ns
or
-bas
ed e
xpl
orat
i
o
n
,
m
o
t
i
on pl
a
n
ni
n
g
, l
o
cal
i
zat
i
o
n
an
d
map
p
i
ng
[1
, 2, 3
,
4
,
5
,
6, 7
]
.
Th
e literatu
re
sh
ows th
at in
t
e
llig
en
t au
ton
o
m
o
u
s
ro
bo
ts are cap
ab
le of dealin
g
wi
t
h
u
n
cert
a
i
n
t
i
es enco
unt
e
r
e
d
i
n
i
t
s
envi
ro
nm
ent
i
n
an i
ndepe
n
d
ent
fa
sh
i
on [
8
,
9,
10
, 1
1
, 1
2
,
1
3
, 1
4
,
4, 1
5
]
.
A
fu
lly au
tonom
o
u
s
r
obo
t h
a
s th
e
p
o
w
e
r to g
a
in in
for
m
at
io
n
abo
u
t
t
h
e
en
v
i
r
o
n
m
en
t th
at can wo
rk
f
o
r
an
ex
tend
ed p
e
riod
with
ou
t hu
man
in
terv
en
tion
[1
6, 1
7
, 1
8
, 19
, 20
].
Su
ch
m
o
b
ile
ro
bo
ts act
au
tono
m
o
u
s
ly
in
di
ffe
re
nt
way
s
suc
h
as an a
u
t
o
n
o
m
ous r
o
bot
`UR
M
A
D
'
pr
o
v
i
d
es assi
st
a
n
c
e
t
o
t
h
e pat
i
e
nt
s i
n
h
o
s
p
i
t
a
l
s
and a
n
aut
o
nom
ous m
obi
l
e
r
o
b
o
t
`M
OV
AI
D'
i
s
i
n
servi
ce t
o
assi
st
t
h
e di
sabl
ed a
nd el
de
rl
y
peo
p
l
e
[2
1]
. R
o
b
o
t
s l
i
k
e
`AB
I
O'
are ca
pabl
e
o
f
sel
f
doc
ki
n
g
t
o
c
h
arge
t
h
ei
r
bat
t
eri
e
s [
2
2]
. T
h
e r
o
b
o
t
l
i
k
e
`
K
he
pe
ra'
i
s
pe
rf
orm
i
ng
aut
o
nom
ous se
rvi
ces i
n
case
of a pa
rt
i
a
l
l
y
kno
w
n
en
vi
r
o
n
m
ent
where
hy
bri
d
m
e
t
hod i
s
used t
o
e
x
pl
o
r
e t
h
e
adva
nt
age
s
o
f
gl
o
b
al
an
d l
o
c
a
l
navi
gat
i
o
n t
a
sks. T
h
e c
o
o
r
di
nat
i
o
n o
f
t
h
e
s
e st
rat
e
gi
es ar
e based
o
n
a f
u
zzy
i
n
fere
nce sy
st
e
m
t
h
at
i
nvol
ve
s on
-l
i
n
e com
p
ari
s
o
n
bet
w
een
t
h
e real
scene
and a
pri
o
r m
e
m
o
ri
zed one [
2
, 23]
.
Th
e
`Seek
u
r' an
d `M
DARS' ro
bo
ts
d
e
m
o
n
s
trate th
eir au
ton
o
m
o
u
s
n
a
v
i
gatio
n
an
d security cap
ab
ilities at an
ai
rbase
[
24]
.
P
r
ot
ot
y
p
e
ur
ba
n
r
o
b
o
t
has
bee
n
devel
o
p
e
d
f
o
r
u
r
ba
n
rec
o
n
n
ai
ssance
m
i
ssi
on
scena
r
i
o
at
Fo
rt
Sam
Ho
u
s
ton
,
with
au
tono
mo
u
s
n
a
v
i
g
a
tion cap
ab
ilities li
k
e
stereo
v
i
sion
-b
ased
ob
stacle av
o
i
d
a
n
ce,
v
i
su
al
serv
oi
n
g
t
o
use
r
-
d
esi
g
nat
e
d
g
o
al
s, a
nd a
u
t
o
n
o
m
ous st
ai
r cl
i
m
bi
ng [1
9]
. T
o
day
aut
o
n
o
m
ous r
o
bot
s a
r
e o
n
hi
gh
dem
a
nd f
o
r
l
a
bo
ri
o
u
s
j
o
b
s
l
i
k
e
dom
est
i
c
cho
r
es,
l
a
u
n
d
r
y
ha
ndl
i
n
g,
cl
ea
ni
n
g
a
n
d
at
t
e
n
d
i
n
g el
de
rl
y
p
e
rso
n
s,
et
c. [
1
5
,
25
,
2
6
,
2
7
,
2
8
,
2
9
,
30]
.
I
n
t
e
rest
i
n
gl
y
,
t
h
e
m
o
st
dem
a
ndi
ng
m
obi
l
e
r
o
b
o
t
s
a
r
e
re
qui
r
e
d
f
o
r
i
n
d
o
o
r
appl
i
cat
i
o
ns.
I
n
or
der
t
o
see
t
h
e hi
gh
dem
a
nd
o
f
se
rvi
ce
r
o
b
o
t
s
t
h
e
re
vi
e
w
i
s
i
n
t
e
nde
d t
o
e
xpl
ore m
o
r
e
st
at
e-
o
f
-th
e
-art techn
o
l
o
g
y
on
m
o
bile rob
o
t
s em
ph
asizing
o
n
th
e em
erg
i
n
g
area of m
u
ltisen
so
r d
a
ta
fu
si
o
n
.
3.
M
U
LTI SENSOR
DA
TA
FU
SION
AN
D IN
TEGRA
TION
To ex
pl
o
r
e t
h
e u
n
k
n
o
w
n
or
part
i
a
l
l
y
kno
w
n
en
vi
r
o
n
m
ent
,
m
obi
l
e
ro
bot
nee
d
s
t
o
m
a
p t
h
e
en
v
i
ron
m
en
t an
d
t
o
m
a
in
tai
n
th
e lo
calizatio
n
p
a
ram
e
ters. For m
o
b
ile robo
t
m
a
p
p
i
ng, th
e rst sign
i
f
ican
t
assignm
ent is to access the ra
nge inform
ation and second
leading assignment is to
convert the ra
nge readi
n
g
in
to
in
tern
al
rep
r
esen
tatio
n. Th
e
robo
t req
u
i
res th
e i
n
tern
al
in
fo
rm
atio
n
to
u
p
d
a
te its state as it m
o
v
e
s aro
und
.
It h
e
l
p
s th
e mo
b
ile
robo
t to
attain
fu
ll au
ton
o
m
y so
th
at i
t
m
a
y o
p
e
rate
with
ou
t
h
u
m
an
in
terv
en
e. It
is an
ex
trem
ely d
i
fa
u
lt task
fo
r mo
b
ile ro
bo
t to tak
e
th
e d
eci
sio
n
with
ou
t up
d
a
ting
th
e
p
r
ev
iou
s
statu
s
o
f
t
h
e
envi
ro
nm
ent
as t
h
e en
vi
r
o
n
m
ent
m
a
y
be hi
g
h
l
y
dy
na
m
i
c. In suc
h
si
t
u
at
i
ons, t
h
e
m
obi
l
e
ro
b
o
t
sy
st
em
accum
u
lates the local e
nvironmental inform
a
tion a
n
d
u
p
dat
e
r
e
cu
r
s
i
v
ely by f
u
si
o
n
pro
cess.
3.
1.
M
u
l
t
i
s
en
sor F
u
si
o
n
During the las
t
decade
,
signifi
cant resea
r
c
h
has m
a
de to solve t
h
e proble
m
s concerni
ng
how t
o
com
b
i
n
e or f
u
se dat
a
fr
om
m
u
lt
i
p
l
e
sourc
e
s i
n
or
der t
o
sup
p
o
rt
deci
s
i
on
-m
aki
ng [1
, 31
, 3
2
]
.
The
t
e
r
m
`in
f
orm
a
t
i
o
n
fu
sion
' b
e
co
mes well establish
e
d
for eng
i
n
e
ering
,
m
e
d
i
cal and
m
i
litary an
d
rob
o
tics
appl
i
cat
i
o
ns, et
c. we
ha
ve
p
r
e
s
ent
e
d
he
re s
o
m
e
im
port
a
nt
defi
ni
t
i
ons
o
f
m
u
lt
i
s
ensor
da
t
a
fusi
on
avai
l
a
bl
e i
n
th
e literatu
re as g
i
ven b
e
l
o
w:
Jo
in
t
Direct
o
r
s of Labo
rato
ri
es (1
987
), defi
n
e
d d
a
ta
fu
sion
as a pro
cess
d
ealin
g
with
th
e asso
ciatio
n
,
co
rrelatio
n
,
co
m
b
in
atio
n
of d
a
ta and in
fo
rm
atio
n
fro
m
sin
g
l
e an
d
m
u
ltip
le so
urces to
ach
iev
e
refi
n
e
d
po
sition
and id
en
tity estimates, an
d
com
p
le
te an
d
timely assess
men
t
s of situ
ati
o
n
s
and
threats, and their significa
nc
e. The
process
is characterized
by
co
nt
i
n
u
ous
refi
nem
e
nts of i
t
s
est
i
m
a
tes and
assessm
en
ts, an
d
ev
alu
a
tion
o
f
th
e
n
eed
fo
r ad
d
ition
a
l so
urces,
o
r
m
o
d
i
ficatio
n
o
f
t
h
e pro
cess itself, to
achi
e
ve i
m
pro
v
ed
resul
t
s
[
1
6]
. In
(1
9
8
7
)
,
Du
rra
nt
-
W
hy
t
e
defi
ne
d f
u
si
o
n
as “The
bas
i
c pro
b
l
e
m
i
n
m
u
lt
i
-
sens
or sy
st
em
s i
s
t
o
i
n
t
e
grat
e
a seque
nce
of
obse
r
vat
i
o
n
s
f
r
om
a num
ber of
di
ffe
re
nt
sens
ors i
n
t
o
a s
i
ngl
e
b
e
st-estim
ate o
f
th
e state of t
h
e env
i
ro
n
m
en
t" [3
3
]
.
Luo
i
n
(199
0),
d
e
fin
e
d
“M
u
ltisen
sor fu
sion
,
refers
to
an
y
stage in a
n
integration proc
ess where t
h
ere is an act
ual
com
b
i
n
at
i
on
(o
r f
u
si
o
n
)
o
f
di
ffe
re
nt
so
ur
ces o
f
sens
ory
i
n
f
o
r
m
at
i
on i
n
t
o
o
n
e re
prese
n
t
a
t
i
onal
f
o
rm
at
" [34]
.
Hal
l
et
al., (
1
9
9
7
),
defi
ned t
h
e “
D
at
a
fusi
o
n
tech
n
i
qu
es com
b
in
e d
a
ta fro
m
m
u
ltip
le s
e
n
s
o
r
s, and
related
in
form
at
io
n
fro
m
asso
ciated
d
a
tab
a
ses, to
achieve im
proved accuracy a
n
d m
o
re speci
fic infere
nces t
h
an could
be a
c
hiev
e
d
by
the use of
a
single sens
or
al
one"
[3
5]
. l
a
t
e
r St
ei
n
b
er
g (
1
99
9
)
,
defi
ned t
h
at “Data fusi
on is the
proces
s
o
f
co
m
b
in
ing
d
a
ta
to
refin
e
state
est
i
m
a
t
e
s and pre
d
i
c
t
i
ons"
[
36]
.
In
(
2
0
0
1
)
,
Dasa
rat
h
y
d
e
fi
ne
d t
h
e “
I
n
f
o
r
m
a
ti
on f
u
si
on
enc
o
m
p
asses t
h
e
t
h
eo
ry
, t
ech
ni
q
u
es, a
n
d t
o
ol
s
concei
ved
an
d
em
pl
oy
ed
fo
r e
xpl
oi
t
i
ng t
h
e s
y
ner
g
y
i
n
t
h
e i
n
f
o
rm
at
i
on acq
ui
r
e
d
fro
m
m
u
ltip
le
so
u
r
ces (sen
so
r, d
a
tab
a
ses, in
fo
rm
ati
o
n
g
a
th
ered
b
y
h
u
man
s
etc.) Such
th
at th
e resu
lting
d
ecision
o
r
actio
n
is so
m
e
sen
s
e
b
e
tter
(qu
a
litativ
ely
an
d
qu
an
titativ
ely, in
term
s o
f
accu
r
acy, ro
bu
st
ness and
et
c.) t
h
a
n
wo
ul
d
be p
o
ssi
bl
e, i
f
t
h
ese
so
urce
s
were
use
d
i
ndi
vi
d
u
al
l
y
wi
t
h
o
u
t
suc
h
sy
ner
g
y
expl
oi
t
a
t
i
on”
[3
7]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
RA I
S
SN
:
208
9-4
8
5
6
Mu
ltisen
so
r
Da
ta
Fu
si
o
n
and In
teg
r
a
tio
n for Mob
ile Ro
bo
t
s
: A Review (KS
N
a
g
l
a
)
13
3
Das (
2
0
0
8
),
de
fi
ne
d t
h
e sens
or f
u
si
o
n
i
n
t
o
di
ffere
nt
l
e
vel
s
such as `
h
i
g
h l
e
vel
fusi
on'
.
“hi
gh l
e
vel
f
u
si
on i
s
t
h
e st
udy
of rel
a
t
i
ons
hi
ps am
ong
o
b
ject
s a
nd
event
s
of i
n
t
e
r
e
st
wi
t
h
i
n
a dy
nam
i
c envi
ro
n
m
ent
”
[38]
. Ll
i
n
as i
n
2
009
m
o
d
i
fied th
e d
e
fin
itio
n
o
f
i
n
fo
rm
atio
n
g
i
v
e
n
as “Info
r
m
a
tio
n
fu
sion
is an
i
n
fo
rm
atio
n
pro
cess
d
ealin
g
with
th
e asso
ci
atio
n
,
correlatio
n, an
d co
m
b
in
atio
n of
d
a
ta
an
d inform
atio
n
fro
m
sin
g
l
e
an
d m
u
ltip
le sen
s
o
r
s
or so
u
r
ces t
o
achi
e
ve re
fi
ne
d est
i
m
a
t
e
s of param
e
t
e
rs, charact
eri
s
t
i
c
s,
event
s
, and
be
havi
ou
rs f
o
r o
b
ser
v
e
d
en
tities in
an
o
b
s
erv
e
d
field o
f
v
i
ew. It is so
m
e
t
i
m
es
i
m
p
l
e
m
en
ted
as a Fu
lly Au
t
o
m
a
t
i
c p
r
o
cess o
r
as a
H
u
m
a
n
-
A
i
d
i
ng pr
o
cess for
A
n
alysis an
d
/
or
Decisio
n
Supp
ort” [
1
6
]
.
Fo
r m
o
b
ile rob
o
t
app
licatio
ns, fu
si
o
n
refers to
an
y stage i
n
the
inte
grati
o
n process
where a
n
actual
com
b
i
n
at
i
on o
f
di
f
f
ere
n
t
so
u
r
ces o
f
i
n
f
o
rm
at
i
on t
a
kes pl
a
ce. The com
b
i
n
at
i
on
de
pen
d
s
up
o
n
t
h
e nat
u
re
of
i
n
f
o
rm
at
i
on t
o
be f
u
se
d
at
di
ffere
nt
l
e
vel
s
of
hi
er
archi
cal
m
odel
as show
n
i
n
Fi
gu
re 1
t
h
e di
eren
t
lev
e
ls o
f
in
fo
rm
atio
n
fusion are classi
fied as:
Si
gn
al
-L
evel
Fusi
on
:
I
t
in
clu
d
e
s sign
al enh
a
n
cem
en
t tech
n
i
q
u
e
su
ch
as b
eam
f
o
r
m
in
g
u
s
ing
m
i
cr
o
phone
arrays. Th
e
resu
ltin
g sign
al fro
m
m
u
ltip
le sen
s
ors is
u
s
u
a
ll
y o
f
th
e sam
e
form
as th
e
o
r
i
g
in
al si
g
n
a
l
but
with
a greater q
u
a
lity.
Pi
xel
-
L
e
vel
F
u
si
on
:
It
re
fers
t
o
fusi
on
of t
h
e i
n
fo
rm
at
i
on i
n
t
h
e f
o
rm
of p
i
xel
s
. The se
ns
ors
pr
o
duce s
u
ch
inform
ation in CMOS or CC
D cam
eras. The fuse
d im
ag
e can be create
d
either by the fusion of pi
xel-by-
pi
xel
or
by
t
h
e
fusi
on
o
f
as
soc
i
at
ed l
o
cal
nei
g
hb
o
r
h
o
ods
o
f
p
i
xel
s
i
n
eac
h
o
f
t
h
e i
m
ages.
Feat
ure-L
e
vel
Fusi
o
n
: It is ap
p
licab
le in
d
i
fferen
t areas such
as
m
obi
l
e
r
o
b
o
t
m
a
ppi
n
g
,
pers
o
n
t
r
ac
ki
n
g
an
d
au
to
m
a
tic
sp
eech
recogn
itio
n
.
In
th
is pro
cess, th
e
features a
r
e extrac
ted from
scene and fuse
d with
ot
he
r se
ns
ory
i
n
f
o
rm
at
i
on s
u
c
h
as
m
i
croph
o
n
es,
et
c.
Symbol-Le
v
el
Fusion
: T
h
e
statistical inference can
be
used fo
r sym
b
ol level fusion
whe
r
e fusion
of
sy
m
b
o
l
s is represen
ted in
t
h
e
fo
rm
o
f
con
d
iti
o
n
a
l
prob
ab
ilit
y.
Fig
u
re 1
.
Fu
n
c
t
i
o
n
a
l Diagram
o
f
Mu
ltisen
so
r In
te
g
r
ation
and Fu
si
o
n
(Figu
r
e red
r
awn from
[3
9
])
3.
2.
M
u
l
t
i
s
en
sor I
n
te
gr
ati
o
n
Mu
ltisen
so
r i
n
tegratio
n
is t
h
e syn
e
rg
istic u
s
e of th
e i
n
form
at
io
n
prov
id
ed
b
y
m
u
lt
ip
le sen
s
ing
devices t
o
assist in the accomplishm
en
t of a
task by a system
. The distin
ction
betwee
n integration and fusi
on
serv
es to
sep
a
rate g
e
n
e
ral issu
es invo
lv
ed in
th
e
i
n
t
e
gr
at
i
on o
f
m
u
l
t
i
pl
e sens
o
r
y
d
e
vi
ces at
t
h
e
sy
st
em
architecture [40,
41]. Hiera
r
chical struct
ures of inte
gration are
use
f
ul
for a
n
efficient re
prese
n
tat
i
on
of
di
ffe
re
nt
l
e
vel
s
an
d f
u
si
o
n
n
ode
s i
n
t
h
e ar
chi
t
ect
ur
e. Exa
m
ples are Na
tional Bu
r
eau
o
f
Stand
a
rd
s (N
BS)
sens
ory
an
d c
ont
rol
hi
e
r
arc
h
y
[42]
. Fi
gu
re
1 re
prese
n
t
s
m
u
lt
i
s
ensor i
n
t
e
grat
i
o
n as a
com
posi
t
e
of
basi
c
fun
c
tion
s
.
Elemen
ts o
f
m
u
lti
sen
s
o
r
i
n
tegratio
n are exp
l
ain
e
d
as fo
llo
ws:
Sensor
s
:
A g
r
o
u
p
of se
ns
o
r
s (H
om
oge
ne
ous
or
Het
e
r
o
gene
o
u
s)
pr
o
v
i
di
ng i
n
p
u
t
t
o
t
h
e i
n
t
e
grat
i
o
n
p
r
o
cess. Raw data filterin
g
and
si
g
n
a
l enh
a
nce
m
en
t can
b
e
p
a
rt
o
f
sen
s
ors.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
089
-48
56
IJR
A
V
o
l
.
3,
No
. 2,
J
u
ne 2
0
14:
1
3
1
– 13
8
13
4
Sensor
Model
: Th
e fun
c
tio
n o
f
sen
s
o
r
m
o
d
e
l is to
conver
t
th
e r
a
ng
e in
fo
r
m
atio
n
fr
om th
e sen
s
o
r
of
d
i
fferen
t
m
o
d
a
lities
in
to
co
mm
o
n
rep
r
esen
t
a
tio
n
.
Th
e ran
g
e in
form
atio
n
p
r
ov
id
ed
b
y
the sen
s
or can
b
e
in
t
h
e f
o
rm
of
v
o
l
t
age, c
u
r
r
ent
,
p
u
l
s
e
wi
dt
h
m
odul
at
ed
si
g
n
al
or
si
g
n
al
i
n
t
h
e
f
o
rm
of a
n
i
m
age
[4
3]
.
Sensor Re
gi
st
rati
on
:
It
i
s
si
gni
fi
cant
t
o
p
r
op
o
r
t
i
onat
e
t
h
e uni
que a
nd t
e
m
poral
di
m
e
nsi
o
ns o
f
se
ns
or
i
n
f
o
rm
at
i
on be
fo
re t
h
e
act
ual
fusi
on
p
r
ocess.
Sensor Proces
sing
:
Fusi
o
n
i
s
do
ne at
t
h
e sy
m
bol
, feat
ure,
pi
xel
l
e
vel
,
a
n
d si
g
n
al
l
e
vel
.
If t
h
e
dat
a
fr
o
m
t
h
e se
nso
r
i
s
si
gni
fi
cant
l
y
di
f
f
e
rent
fr
om
ot
h
e
r se
ns
ors,
i
t
c
a
n
be se
parat
e
d
fr
om
t
h
e fusi
on
p
r
ocess.
World Model
:
Duri
n
g
na
vi
gat
i
o
n
,
a
m
o
b
i
l
e
robot
ext
r
act
s t
h
e i
n
for
m
at
i
on from
the sens
or
s an
d
gene
rates the
local
m
a
p with res
p
ect to t
h
e curre
nt
po
sitio
n
.
Th
e info
rm
atio
n
is up
d
a
ted
with
prior
i
n
f
o
rm
at
i
on t
h
at
gene
rat
e
s t
h
e w
o
rl
d
m
odel
.
The
w
o
rl
d m
odel
i
s
us
ual
l
y
defi
ned
i
n
t
e
r
m
s of hi
g
h
l
e
v
e
l
represen
tatio
n
for m
u
ltisen
so
r fu
sion
in m
o
b
ile robo
t n
a
v
i
g
a
tio
n
.
Sensor Selec
t
ion
: It en
ab
les th
e
m
u
ltisen
so
r system to
s
e
lect th
e
m
o
st
ap
p
r
op
riate co
nfigu
r
ation
of
sens
or
[4
4]
. T
h
e se
nso
r
sel
e
c
t
i
on ca
n be cl
a
ssi
fi
ed as:
a)
P
re-Selecti
o
n
: It is th
e p
r
im
ary step
to
ward
s a
gene
ral
m
e
t
hodol
ogy
t
o
sel
e
ct
a sui
t
a
bl
e sens
or i
n
res
p
e
c
t
t
o
envi
ro
n
m
ent
a
l
condi
t
i
ons
. P
r
e-sel
ect
i
o
n
d
e
p
e
nd
s upo
n
g
e
o
m
etric lo
catio
n
of sen
s
ors an
d
sta
tic/d
yna
m
i
c co
n
d
itio
ns o
f
m
o
b
ile rob
o
t
(Ho
v
l
an
d
et
al., 19
97)
.
b
)
Real Time
:
[
4
5]
pres
ent
e
d t
h
e app
r
oach
of
sens
or sel
ect
i
o
n i
n
r
eal
t
i
m
e
by
eval
uat
i
n
g t
h
e
per
f
o
r
m
a
nce v
a
l
u
e o
f
eac
h
s
e
ns
or
[
45]
.
If
t
h
e
per
f
o
r
m
a
nce val
u
e
of a
p
a
rt
i
c
ul
ar se
ns
o
r
i
s
l
o
w,
t
h
e
n
t
h
e
alg
o
rith
m
rej
e
cts th
at sensor t
o
p
a
rticip
ate
for in
teg
r
ation
.
Sys
t
em Contr
o
ller
: Syste
m
co
n
t
r
o
ller
execu
tes th
e comman
d
s
to
the
m
o
b
ile r
obot actu
a
to
r
s
.
Th
e
alg
o
rith
m
lik
e p
a
th
p
l
ann
i
ng
, co
llisio
n
avo
i
d
a
n
ce, and
n
a
v
i
g
a
tio
n
rely u
p
o
n
th
e feed
b
a
ck
sign
al
o
f
sens
ors
.
3
.
3
.
A
d
va
ntages
of
M
u
lt
isenso
r D
a
ta
Fusio
n
For
m
obi
l
e
ro
b
o
t
ap
pl
i
cat
i
ons
Pot
e
nt
i
a
l
a
dva
nt
ages
o
f
m
u
l
t
i
sens
or
dat
a
fu
s
i
on a
r
e
gi
ve
n a
s
:
Reduc
tio
n
o
f
Uncer
tain
ty
:
Sens
or
s p
r
o
v
i
de o
n
l
y
t
h
e e
s
t
i
m
a
ti
on o
f
r
a
nge
w
h
i
c
h m
a
y
be u
n
cert
a
i
n
.
Mu
ltisen
sor
d
a
ta fu
si
o
n
red
u
ces th
e
u
n
c
ertain
ty as th
e fusio
n
p
r
o
cess is redun
d
a
n
t
.
Hen
ce, it in
creases
the accuracy
which t
h
e syste
m
percei
ves from
the envi
ronment [41, 43].
a.
Unce
rt
ai
nt
y
i
n
Sens
ory
In
fo
rm
ati
on:
U
n
ce
rt
ai
nt
y
i
n
t
h
e
sens
ory
i
n
fo
r
m
at
i
on can
be
cause
d by
l
i
m
i
t
e
d resol
u
t
i
on
of t
h
e se
nso
r
,
ran
d
o
m
m
easurem
ent
of
noi
se
, sy
st
em
at
i
c
errors
and
d
u
e t
o
in
co
m
p
leten
e
ss o
f
t
h
e in
formatio
n
e.g
.
Sin
g
l
e
fix
e
d
camera can
no
t sen
s
e th
e en
tire in
fo
rm
atio
n
of
th
e env
i
ro
n
m
en
t du
e to
limit
ed
v
i
ew. To
co
m
p
lete th
e in
form
at
io
n
m
u
ltip
le v
i
ews are n
eed
ed
to
form
the com
p
lete local view
[46].
b.
Un
certain
t
y in th
e Env
i
ro
n
m
en
t: Th
e m
o
b
ile ro
bo
t
e
nvi
ro
nm
ent
becom
e
s u
n
certai
n
,
when no prior
i
n
f
o
rm
at
i
on i
s
avai
l
a
bl
e or t
h
e envi
r
o
nm
ent
i
s
hi
ghl
y
dy
na
m
i
c. The ro
bot
s ope
rat
e
i
n
u
n
d
er
wat
e
r a
n
d
space e
x
plorati
ons
are
hi
ghly
unce
r
tain a
b
out the environment
[28, 47, 48,
49].
c.
Unce
rtainty in Robot L
o
c
a
lization: For accura
te m
a
ppi
ng
robot needs accurate localization
p
a
ram
e
ters su
ch
as, m
o
b
ile ro
bo
t `po
s
itio
n
'
an
d
`o
rien
tati
o
n
'
. Odo
m
etric
erro
rs
du
e to
wh
eel slip,
in
clin
atio
n of
ro
bo
t can
cau
s
e
p
o
s
ition
and
o
r
ien
t
atio
n
erro
rs [5
0
]
.
Com
p
lementary
: Mu
ltisen
sor d
a
ta fu
si
o
n
is a co
m
p
le
m
e
n
t
ary p
r
o
cess
b
e
cau
se it allo
ws p
e
rcei
v
i
ng
the
i
n
f
o
rm
at
i
on o
f
di
ffe
re
nt
pa
rt
s
of
t
h
e e
n
vi
ro
n
m
ent
by
di
ffe
r
e
nt
se
nso
r
s
[
7
]
.
Well-Timed
: Mu
ltisen
sor
d
a
ta fu
si
o
n
i
n
creases th
e
p
r
o
c
essin
g
sp
eed
due to
th
e
pro
cess of p
a
rallelism
[3
7]
.
Less Cos
t
ly
:
Si
ngl
e sen
s
or
needs se
v
e
ral
el
ect
roni
c
m
odul
es t
o
pr
ocess t
h
e
si
gnal
,
w
h
e
r
eas
com
m
onproces
sing m
o
dule of m
u
ltisenso
r data
fusion process reduces th
e overall c
o
s
t
of the
system
[5
1]
.
Increase
d
Confidence and Reduce
d
Am
biguity
: If sev
e
ral sen
s
ors con
t
ribu
te to
a m
e
a
s
u
r
em
en
t resu
l
t
,
t
h
e l
e
vel
of
co
nfi
d
ence
o
f
t
h
e
f
u
sed
val
u
e
be
com
e
hi
ghe
r [
5
2]
Increase
d Reliability
:
A sy
st
em
rel
y
i
ng
on
di
ffe
re
nt
se
nso
r
s i
s
l
e
ss
s
u
sce
p
t
i
b
l
e
t
o
di
st
ur
bance
ca
used
b
y
hum
an act
i
o
ns
or
nat
u
ral
phe
n
o
m
e
na [5
3]
.
Enhanc
ed S
p
a
ti
al Res
o
lu
tion
: M
u
ltip
le sen
s
o
r
s d
a
ta
fusio
n
en
ab
les
th
e syste
m
to
enhan
ce an
d in
crease
t
h
e m
a
p resol
u
t
i
on
[5
4]
.
3.
4. Mul
t
i
Sen
s
or Da
t
a
F
u
si
on Al
g
o
ri
t
h
ms
In
th
is sectio
n o
u
r
rev
i
ew is in
ten
d
e
d
to
fin
d
v
a
riou
s m
e
th
od
s u
s
ed
to
fu
se th
e informatio
n
for
m
a
ppi
n
g
a
n
d l
o
cal
i
zat
i
on.T
h
e dat
a
f
u
si
o
n
m
e
t
hods
can
b
e
cl
assi
fi
ed
as
Est
i
m
at
i
on M
e
t
h
o
d
(t
hat
i
n
c
l
udes
recu
rsive
an
d
n
o
n
-rec
u
rsive m
e
tho
d
)
,
Classifi
cati
on M
e
t
h
o
d
,
I
n
fe
rence
M
e
t
h
o
d
, and
Artificial Meth
o
d
.
Wei
g
h
t
ed A
v
erage
Me
th
od
of m
u
ltisensor data fusion i
s
th
e res
p
onsi
ve and sim
p
le
m
e
thod in
whi
c
h a
wei
g
ht
ed a
v
era
g
e
of
r
e
du
n
d
ant
i
n
f
o
r
m
at
i
on pr
o
v
i
d
e
d
by
a g
r
o
u
p
o
f
se
nso
r
s i
s
use
d
as t
h
e f
u
se
d
val
u
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
RA I
S
SN
:
208
9-4
8
5
6
Mu
ltisen
so
r
Da
ta
Fu
si
o
n
and In
teg
r
a
tio
n for Mob
ile Ro
bo
t
s
: A Review (KS
N
a
g
l
a
)
13
5
[1
6]
. A wei
g
ht
ed aver
age i
s
u
s
ed i
n
va
ri
o
u
s m
obi
l
e
robot
s
suc
h
as “HIL
A
R
E" i
n
whi
c
h t
h
e i
n
f
o
rm
at
i
on from
m
u
lt
i
p
l
e
sens
o
r
s i
s
fu
sed
by
usi
n
g
wei
g
ht
e
d
a
v
era
g
e m
e
tho
d
[
52]
.
Thi
s
m
e
t
hod
i
s
not
sui
t
a
bl
e
f
o
r
d
y
n
am
i
c
envi
ronm
ent as com
p
ared to t
h
e static environm
ents.
Ka
lma
n
Filter
is a set of
m
a
them
atical equations that pr
ovi
d
es an e
fficient com
putational
m
eans to
estim
a
te
the state of a
proces
s in a
way, th
at, it
min
i
m
i
zes
th
e m
ean
o
f
the sq
uared
error [55
]
. Jetto
(199
9) in
hi
s resea
r
c
h
u
s
ed a
n
e
x
t
e
n
d
e
d
Kal
m
an fi
l
t
er t
o
fu
se i
n
fo
rm
ati
on o
f
e
n
code
rs a
n
d s
o
nar
sens
o
r
s [
5
6]
. A
n
ex
tend
ed
Kalman
filter is
u
s
ed
to so
lv
e t
h
e
co
n
c
urre
n
t
m
a
p
p
i
n
g
and lo
calizatio
n
(CML)
o
f
th
e m
o
b
ile ro
bo
t
[57]. Recently
, exte
nde
d
Ka
l
m
an filter is use
d
to c
o
m
b
ine ultrasonic and
stere
o
ca
mera inform
at
ion t
o
increase the robust
n
ess of the
m
a
p [58].
'Extended Ka
lman
Filter'
(EKF) served as the
prim
ary approach to
m
a
p dy
nam
i
c
envi
ro
nm
ent
for t
h
e l
a
st
seve
ral
y
ears but
i
t
suf
f
ers
fr
om
two
wel
l
kn
o
w
n sh
ort
c
om
i
ngs. These
two
p
r
o
b
l
em
s
are th
e qu
ad
rat
i
c co
m
p
lex
ity,
an
d
t
h
e sen
s
iti
v
ity to
failu
res in
d
a
ta asso
ci
atio
n
[5
1
]
. Th
e EKF
has
bec
o
m
e
wi
del
y
kn
o
w
n
i
n
t
e
rm
s of
g
r
owt
h
of
com
p
l
e
xi
t
y
due t
o
t
h
e
up
dat
e
st
e
p
t
h
at
re
q
u
i
r
es
l
a
rge
com
put
at
i
on t
i
m
e
pr
op
ort
i
o
na
l
t
o
t
h
e
sq
uar
e
of
t
h
e
n
u
m
b
er of
l
a
n
d
m
a
rks i
n
t
h
e
en
vi
r
o
nm
ent
.
Demps
t
er-Shafer
,
(DS) t
h
eory of m
u
ltisen
so
r d
a
ta
fusion is u
s
ed
t
o
redu
ce th
e un
certain
t
y in
th
e
gri
d
cau
sed
d
u
e
t
o
t
h
e se
ns
or
y
i
n
fo
rm
at
i
on whe
r
e t
h
e
wei
ght
of
co
nfl
i
c
t
m
e
t
r
i
c
and t
h
e enl
a
r
g
em
ent
of t
h
e
fram
e
of di
scernm
ent
are t
h
e t
w
o c
o
m
pone
nt
s use
d
t
o
m
easure t
h
e am
ount
o
f
cons
ens
u
s bet
w
een di
erent sens
ors. Lack of
c
ons
e
n
sus
lead
s th
e
robo
t to
eit
h
er
co
m
p
en
sate wi
th
in
certain
li
mits o
r
inv
e
sti
g
ate th
e
p
r
ob
lem
fu
rth
e
r; with th
is it
help
s in add
i
ng
ro
b
u
st
ness
t
o
t
h
e
ro
b
o
t
'
s ope
r
a
t
i
on
[2
7,
5
9
]
.
Artificial Neu
r
al
Ne
tw
ork
s
use
d
t
o
m
a
p t
h
e occ
upa
ncy
g
r
i
d
has
p
r
o
v
e
n
t
o
b
e
r
o
bust
a
n
d
ada
p
t
i
v
e t
o
t
h
e en
vi
r
onm
ent
a
l
chan
ges
[
6
0
,
6
1
,
6
2
, 6
3
,
64]
.
[6
5]
p
r
o
p
o
sed
bac
k
-
p
r
o
pagat
i
o
n t
r
ai
ni
ng
o
f
m
u
l
t
i
-
l
a
y
e
r
perce
p
t
i
o
n [6
5
]
. The neu
r
al
n
e
t
w
o
r
k i
s
t
r
ai
n
e
d t
o
pe
rf
orm
t
h
e cor
r
ect
con
v
ersi
on
of ra
n
g
e i
n
f
o
rm
at
i
on i
n
t
o
occu
pa
ncy
gri
d
. I
n
t
h
e w
o
r
k
of T
h
r
u
n, t
h
e ro
b
o
t
obt
ai
ns t
h
e t
r
ai
ni
n
g
sam
p
l
e
s by dri
v
i
ng ar
o
u
nd i
n
a
cal
i
b
rat
i
on e
n
vi
r
onm
ent
[2
8
]
. Dam
(199
6)
i
n
hi
s pa
per
pr
o
pose
d
a ne
ural
net
w
o
r
k m
e
t
hod t
o
l
e
a
r
n t
h
e
p
r
ob
ab
ilistic so
n
a
r sen
s
or mo
d
e
l. Th
e
co
nversion
of th
e sen
s
o
r
d
a
ta remain
s ad
ap
tiv
e to
ch
ang
e
in
eith
er the
sen
s
o
r
o
r
its en
v
i
r
o
n
m
en
t [
6
6
]
. K
a
m
(
1
997)
pr
esen
ts a h
i
er
ar
ch
ical n
e
u
r
al n
e
tw
ork
fo
r
m
o
b
ile r
o
bo
t co
n
t
r
o
l
.
The
network receives input
from
the sens
ors
and tr
a
n
s
m
its on/off commands to the m
o
tors. B
u
t
m
a
jor
d
r
awb
ack is t
h
at larg
e tim
e
is requ
ired
to train th
e
net
w
o
r
k
[
6
7]
Lat
e
r l
a
r
g
e
w
o
r
k
i
s
d
one
on
N
N
by
di
erent researche
r
s
[68, 69, 64].
H
i
s
t
og
ra
m
m
i
c in
Mot
i
o
n
M
a
p
p
i
ng
,
(
H
I
M
M
)
al
g
o
ri
t
h
m
devel
ope
d
by
B
o
ren
s
t
e
i
n
an
d
Ko
re
n i
n
19
9
1
at
t
h
e
Un
i
v
ersi
t
y
of M
i
c
h
i
g
a
n
w
h
i
c
h
p
r
ovi
des a
di
ffe
r
e
nt
ap
pr
oac
h
t
o
sc
ore
whet
he
r a pa
rt
i
c
ul
ar el
em
ent
in
an
o
ccup
a
n
c
y g
r
id
is
o
ccupied
or
em
p
t
y [
7
0, 43
]. T
h
e main objecti
v
e of
H
I
M
M
was t
o
i
m
prove
o
b
st
acl
e
avoi
dance
f
o
r
m
obi
l
e
rob
o
t
.
Bayesian
Meth
od
allows m
u
ltisen
sory info
rm
atio
n
to
be co
m
b
in
ed
acco
rd
ing
t
o
th
e ru
les
o
f
p
r
ob
ab
ility th
eo
ry. Bayes' ru
l
e
of co
m
b
in
ati
o
n allo
ws
th
e
co
m
b
in
in
g of
a prio
ri
prob
abilit
y o
f
a
h
ypoth
e
sis
wi
t
h
t
h
e co
ndi
t
i
onal
pr
o
b
abi
l
i
t
y
of gi
ven h
y
pot
he
si
s [7
1,
72
, 7
3
, 7
4
]
.
M
o
ra
vec (
1
9
8
5
)
at
C
a
rnegi
e
M
e
l
l
o
n
Un
i
v
ersity p
i
on
eered
th
e
p
r
ob
ab
ilistic ap
pro
ach. Later M
o
ra
v
ec turn
ed
i
n
to
a form
o
f
Bayes'
Ru
le wh
ich
u
s
es
p
r
ob
ab
ilities ex
pressed
as likelih
o
o
d
s
and
od
d
s
[57
]
.
Fu
zzy Log
i
c based
sen
s
or fu
sio
n
relates to
t
h
e artificial in
tellig
en
ce class o
f
m
u
ltisen
so
r
d
a
ta fusion
.
Th
is m
e
th
o
d
can
also
b
e
consid
ered
as a
possib
ilistic ap
pro
ach in th
e sen
s
e t
h
at th
e meth
od
do
es
no
t assign
p
r
ob
ab
ilities to
th
e
p
r
op
ositio
n
s
b
u
t
it assig
n
s
th
e m
e
m
b
ersh
ip
v
a
lues to
pro
p
o
s
itio
n
[39
]
. Th
ere is
trem
endous
fle
x
ibility to perform
fusion
of
m
u
ltisensory i
n
form
ation under
t
h
e special
rule
of c
o
m
b
ination of
fuzzy values
.
4.
CO
NCL
USI
O
N
In m
o
b
ile robo
ts th
ere are ch
allen
g
e
s to
dev
e
lop
b
e
tter an
d effien
t
syste
m
to
to wo
rk in
co
m
p
lex
envi
ro
nm
ent
s
. Li
t
e
rat
u
re
sh
ows
t
h
at
t
h
er
e i
s
am
pl
e scope
f
o
r
de
vi
si
ng i
m
pl
em
ent
a
t
i
ons
i
n
e
x
i
s
t
i
n
g
m
u
lt
i
s
ensor
da
t
a
fusi
o
n
fram
e
wo
r
k
s. B
a
y
e
s
i
an i
s
t
h
e ol
de
st
app
r
oac
h
a
n
d o
n
e
wi
t
h
st
r
o
n
g
est
f
o
un
dat
i
on.
Bayesian
and
DS m
e
th
o
d
s
hav
e
so
m
e
fund
am
en
tal p
r
oble
m
s lik
e in
form
at
io
n
u
n
c
ert
a
in
ty, con
f
licts and
in
co
m
p
leten
e
ss. Sen
s
or
fu
sio
n
u
s
ing
N
N
r
e
qu
ir
es long ti
m
e
to
tr
ain
th
e
m
o
b
ile ro
bo
t fo
r
a
p
a
r
ticu
l
ar
envi
ro
nm
ent
and i
t
i
s
co
nsi
d
ered as
di
f
f
i
c
ul
t
for c
o
m
p
lex
en
v
i
ron
m
en
t with
larg
e v
a
riatio
n
s
ex
ists. HIMM is
li
mited
to
son
a
r, bu
t it h
a
s si
gn
ifican
t co
m
p
u
t
atio
n
a
l ad
vant
age. In practic
e
Bayesi
an
m
e
th
od
of informatio
n
fusi
on
i
s
f
o
u
n
d
t
o
be m
o
re
st
rai
g
ht
fo
r
w
a
r
d t
o
a
d
o
p
t
f
o
r i
n
do
o
r
an
d
out
do
o
r
en
vi
r
onm
ent
.
To m
a
ke t
h
e
m
a
ppi
n
g
a
n
d
l
o
cal
i
zat
i
on
ro
b
u
st
t
h
ere i
s
nee
d
o
f
pr
epr
o
cessi
ng a
nd
p
o
st
p
r
o
c
e
ssi
ng
o
f
t
h
e
sens
ory
in
fo
rm
atio
n
and
resu
ltan
t
i
n
tern
al
represen
tatio
n
in th
e
fo
rm
o
f
m
a
p
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
089
-48
56
IJR
A
V
o
l
.
3,
No
. 2,
J
u
ne 2
0
14:
1
3
1
– 13
8
13
6
REFERE
NC
ES
[1]
M Asada, Y Fukui,
and S Tsuji.
“Rep
resenting g
l
obal world of
a
mobile
robot with relation
a
l lo
cal maps”.
Sy
ste
m
s,
Man and Cyb
e
rn
etics, I
EEE Transactions on
. 199
0; 20(6): 1456-1
461.
[2]
W Yaonan, Y
Yimin, Y Xiao
f
a
ng,
Z Yi, Z Yuanli, Y Feng,
and T Lei.
“Autonomous mobil
e
robot navig
a
tion
s
y
stem designed
in dy
n
a
mic env
i
ronment
based on transferable belief model”.
Measurement
. 2
011; 44(8): 1389-
1405.
[3]
J Andrade-Cetto
and A Sanf
eliu
. “Concurren
t
map build
i
ng
and
localization on
indoor
d
y
n
a
mic env
i
ronments”.
International
Jo
urnal of
Patt
ern
R
ecogn
ition
and
Artif
icial In
telli
gence
. 2002
; 16(
03): 361-374.
[4]
S Garrido, L Moreno, and D Blanco. “Explor
ati
on and mapping using the vfm
motion planner”.
Instrume
ntation
and Measurement, I
E
EE Transactions on
. 2009
;
58(8) 2880-2892
.
[5]
S Blavzivc. “A novel trajector
y
-tr
ack
ing c
ontr
o
l law for wheeled mobile rob
o
ts”.
Robotics and Autonomous
Systems
. 2011; 5
9
(11): 1001-100
7.
[6]
M Knudson and
K Tumer. “Adap
tive
navig
a
tion
f
o
r autonomous r
obots”.
Robotics and Au
tonomous Systems
. 2011;
59(6): 410-420
.
[7]
K Hou, H Sun, Q Jia, and Y
Zhang. “
An aut
onomous positioning and navigation
system for spherical mobile
robot
”. Procedia Engin
eering
.
20
12; 29: 2556-25
61.
[8]
O Bozma and R Kuc. “Buildin
g a sonar map in a sp
ecular en
vironment using a
single mobile sensor”.
Pat
t
er
n
Analysis and
Ma
chine
Inte
llig
enc
e
, I
E
E
E
Transactions on
. 1991; 1
3
(12): 1260-126
9.
[9]
P Gaudiano,
E
Zalama, and
JL Corona
do. “An unsupervised n
e
ural n
e
twork f
o
r low-lev
e
l co
ntrol of
a wheeled
mobile robot: noise resistance,
stab
ili
t
y
, and ha
rdware im
plem
entation
”
.
Systems, Man, and Cybe
rnetics, Part B:
Cybernetics, IEEE Transactions
on
. 1996; 26(3)
: 485-496.
[10]
G Anousaki an
d K K
y
riakopo
ulos. “Sim
ultan
e
ous localizatio
n and map build
ing for mobile robot nav
i
gatio
n”.
Robotics
&
Auto
mation Magazin
e, I
E
EE
. 1999; 6
(
3): 42-53.
[11]
D Schulz and
W Burgard. “Probabilisti
c state
esti
m
a
ti
on of d
y
nam
i
c objects with
a m
oving m
obil
e
robot
”.
Robotics
and Autonomous
Systems
. 2001;
34(2): 107-115
.
[12]
P Goel and GS Sukhatme. “
Sonar-based featu
r
e recognition a
nd robot
navigation using a neural network
”. i
n
Intell
igen
t Robots and Sy
st
em
s, 2000. (IROS 20
00). Proceedin
gs. 2000 IEEE/
RSJ Internationa
l Conferenc
e
on
.
2000; 1: 109-11
4.
[13]
S Atiy
a and GD
Hager. “Rea
l-time vision-based
robot lo
cal
ization”. Robotics an
d Auto
mation, I
EEE
Transactio
ns
on. 1993; 9(6): 7
85-800.
[14]
H Durrant-W
hyte and T Bail
e
y
. “
S
im
ulta
neo
u
s localization and mapping: part-I”.
Robotics &
Automation
Magazine, I
E
EE
. 2006; 13(2): 99
-110.
[15]
B Lau, C Sprunk, and W Burgard. “Efficient gr
id-based
spatial representations
f
o
r
robot navigation in d
y
n
a
mic
environments”.
Robotics
and Au
tonomous Systems
. 2012.
[16]
J Llin
as, DL
Hal
l
,
and ME
Ligg
i
n
s. Handbook of
Multisenso
r Dat
a
Fusion:
Theor
y
and Pract
i
ce.
CRC Press, 200
9.
[17]
HC Lai,
R Yang,
and GW Ng.
“
Enhanced self-organizing map
for passive
son
a
r tracking to improve situatio
n
awareness
”. in
I
n
formation Fusion, 2007
10th In
tern
ational Conf
erence on
. I
EEE. 2007: 1-7
.
[18]
A Birk, N Vaskevicius, K Path
ak, S Schwertf
eger, J
Poppinga, and H Bulow.
“3-d
perception
and modeling”.
Robotics
&
Auto
mation Magazin
e, I
E
EE
. 2009; 1
6
(4): 53-60.
[19]
L Matth
ies, Y X
i
ong, R Hogg, D
Zhu, A
Rank
in, B Kenned
y
, M
Hebert, R Maclac
hlan
, C Won, T Frost, et
al. “A
portable, autono
mous, urba
n reconnaissance rob
o
t”.
Robotics an
d Autonomous S
y
stems
. 2002; 40
(2): 163-172
.
[20]
RC Luo and C
C
Lai
.
“Enrich
e
d indoor m
a
p construction
b
a
sed on m
u
ltisensor fusion approach for int
e
ll
igen
t
service robot”.
I
ndustrial Elec
tronics, I
E
EE Transactions on
. 201
2; 59(8): 3135-3
145.
[21]
P Dario, E Guglielmelli, V Genove
se, and M Toro. “Robot assi
stants
: Applications and evolutio
n”.
Robotics and
Autonomous Sys
t
ems
. 1996; 18(1
)
: 225-234.
[22]
SK Chalup, CL Murch, and MJ
Quinlan.
“Machine learn
i
ng with
aibo robots in the fourlegged league of robocup
”.
Systems, Man
,
a
nd Cybernetics
,
Part C:
A
pplications and Review
s,
IEEE Transactions on
. 2007; 3
7
(3): 297-310
.
[23]
H Maaref
and C
Barret. “Sensor-based fuzzy
nav
i
gation of
an
auto
nomous mobile robot
in
an indoo
r environment”.
Control Engin
e
ering Practice
. 20
00; 8(7): 757-76
8.
[24]
D Bogdan, F Adrian, M Viorel,
V Alina, and
M Eugenia. “
Di
screte-t
ime slidi
ng-mode
control of four driving
-
steering wh
eels
autonomous vehicle
”. in Con
t
rol
Conference (CC
C
),
2011
30th C
h
inese. IEEE. 20
11: 3620-3625.
[25]
H Moravec and A Elfes. “
High resolution maps from wide angle sonar
”. in Robotics and Auto
mation. Proceeding
s
.
1985 IEEE In
ter
n
ation
a
l Conf
erence on
. 1985
; 2
:
116-121.
[26]
A Elfes. “Using
occupan
c
y
gr
ids
for m
obile
robot percep
tion and
n
a
vigation”.
Com
puter
. 1989
; 22(
6): 46-57.
[27]
R Araujo
and
AT de Almeid
a. “Learning
sensor-based
nav
i
gation of a real m
obile robo
t in
u
nknown worlds”.
Sy
ste
m
s,
Man,
and Cy
be
rnetic
s
,
Part
B: Cy
be
rnetic
s,
IEEE Transactions on
. 1999
; 29(2): 164-178.
[28]
S Thrun. “Learn
i
ng occup
a
ncy
gr
id ma
ps with
for
w
ard
sensor models”.
Au
tonomous robots
. 2003;
15(2): 111-127
.
[29]
RR Murph
y
. “Dempster-shafer
theor
y
for sens
or
fusion in
auton
o
mous mobile robots”.
Robotics and Au
tomation
,
IEEE Transactio
ns on
. 1998; 14(
2): 197-206.
[30]
R Siegwart and
I
R
Nourbakhsh. I
n
troduction
to
Au
tonomous Mob
ile Robotos. The MIT pr
ess. 201
1.
[31]
D Blanco, B Boada, L M
o
ren
o
, and M
S
a
lichs
. “
Local map
p
ing from online laser Voronoi extraction
”. in
Intelligen
t Robo
ts and S
y
stems,
2000.(IROS 20
00). Proceeding
s
. 2000 I
EEE/R
SJ International Conferen
ce on.
2000; 1: 103-10
8.
[32]
JJ Leonard, HF Durrant-Wh
y
te, and IJ
Cox. “D
y
n
amic map building fo
r an
autonomous
mobile robot”.
The
International Jo
urnal of
Robotics Research
. 199
2; 11(4): 286-29
8.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
RA I
S
SN
:
208
9-4
8
5
6
Mu
ltisen
so
r
Da
ta
Fu
si
o
n
and In
teg
r
a
tio
n for Mob
ile Ro
bo
t
s
: A Review (KS
N
a
g
l
a
)
13
7
[33]
HF Durrant-Wh
y
t
e.
Integr
ation
,
coor
dination
and control
of
m
u
lti-sensor
robot s
y
st
em
s. Kluwer Acad
em
ic
Publishers, 1987
.
[34]
RC Luo and MG Kay
.
“
A tu
torial on multisens
o
r integration a
nd fusion
”. in In
dustrial E
l
e
c
tron
ics Societ
y,
199
0.
IECON'
90., 16
th
Annual Conf
eren
ce of
IEEE. I
E
EE.
1990: 707-7
22.
[35]
DL Hall and J Llinas.
“
An
introd
uction
to mul
tise
n
sor data fusion
”. Proceed
ings o
f
the IEEE. 1997
; 85(1): 6-23.
[36]
AN Steinberg, CL Bowman, and FE White. “Revisions
to the jdl data fus
i
on m
odel”. in
AeroSense'99.
International So
ciety
for Optics
and Photon
ics.
1
999: 430-441.
[37]
BV Dasarath
y
.
“
I
nformation fusion-wh
at, wh
ere
,
wh
y
,
when,
and
how
?
"
.
Information Fusion
. 2001
; 2(2): 75-76.
[38]
S Das. High-lev
e
l d
a
ta fusion. A
r
tech
House, 200
8.
[39]
A Plascencia. Sensor fusion for autonomous mobile robot
nav
i
gation
.
Videnbasen for Aalborg UniversitetVB
N
,
Aalborg Univer
sitet Aalborg Universit
y
, De
t T
e
knisk-Nat
urvid
e
nskabelige Fak
u
ltet Th
e Faculty
of Engineerin
g
and Science, Automation & Cont
rolAutomation
& Control, 2007
.
[40]
B A
y
rulu
, B
B
a
rshan, I
Erkm
en,
and A
Erk
e
n. “
E
vid
e
nt
ial
logic
a
l
sensin
g using m
u
ltipl
e
sonars for
th
e
identif
ication
of
targ
et pr
im
itives in a m
obil
e
r
obot'
s
environm
ent”.
in
Multisensor Fusion and Integration
fo
r
Intelligen
t Systems, 1996. I
E
EE/SICE/
RS
J International Con
f
eren
ce on
. I
E
EE
. 199
6: 365-372.
[41]
RC Luo, CC Yih, and KL Su. “Multisensor fusion and integr
a
t
i
on: approach
es,
appli
cat
ions, an
d future resear
c
h
directions”.
Sens
ors Journal, I
E
EE
. 2002
; 2(2)
: 1
07-119.
[42]
DDL Hall
and S
AH McMullen.
Mathemat
ical techniques
in multisensor da
ta fusion. Artech Hous
e. 2004
.
[43]
R Murph
y
. An
introduction
to
AI robotics.
The
MIT press. 2004
.
[44]
T Koshizen
. “Improved sensor selection
techn
i
que b
y
in
tegr
ating sensor fusio
n
in
robot p
o
sition estimation”.
Journal of Intelligent and
Robotic Systems
. 2000; 29(1): 79-9
2
.
[45]
O Cohen and
Y Edan. “A sensor fusion framework
for online sensor and
algorithm selection”.
Robotics and
Autonomous Sys
t
ems
. 2008; 56(9
)
: 762-776.
[46]
SJ Henkind and MC Harrison. “An analy
s
is of
four uncer
ta
int
y
c
a
lcu
li”
. S
y
st
em
s, Man and C
y
bern
et
ics, IE
EE
Transactions on
. 1988; 18(5)
: 70
0-714.
[47]
A Elfes. “Sonar-based real-w
orld
mapping and navigation
”
.
Robo
t
i
cs and Automati
on, IEEE Journa
l of.
1987; 3(3):
249-265.
[48]
RC Luo and
MG Ka
y. “
M
ultise
n
sor integr
ation
and fusion in
int
e
llig
ent s
y
stem
s”.
S
y
stems, Man
and Cybernetics
,
IEEE Transactio
ns on
. 1989; 19(
5): 901-931.
[49]
G Brooker
and
AT Brooker
.
“In
t
roduction
to
sen
s
ors for ra
nging
and imaging
”
. `SciTech Pub
.
In
co
rporated
, 2009
.
[50]
G Lawitzk
y
, W Feiten
,
and M Moller. “Sona
r sensing for low-cost indoor m
obilit
y
”
.
Robo
tics and
Autonomous
Systems
. 1995; 1
4
(2): 149-157
.
[51]
J Bi
j
k
e
r
a
nd W
St
ey
n.
“Ka
l
ma
n
fol
t
e
r c
onfigur
ations for a
low-co
st loosely
in
t
e
gr
ated
iner
tia
l n
a
vi
gation
s
y
st
em
on
an airship”.
Control Engin
eering
Practice
. 2008;
16(12): 1509-15
18.
[52]
M Khatib and T Simeon. “
Senso
r
-based motion planning and co
ntrol for the hilare mobile robot
”. in Int
e
l
ligen
t
Robots and S
y
st
em
s, 1997. IROS'
97., Proceed
in
gs of the
1997 IEEE/RSJ Intern
a
tional Conf
eren
c
e
on. 1997; 3: V
8
-
V9.
[53]
HM Choset. Prin
ciples of robo
t motion:
theor
y
,
algorithms, and
implemen
tations.
MIT press, 2005
.
[54]
D Macii, A B
oni, M De Ce
cco,
and D Pe
tri.
“
T
utor
ia
l 1
4
: Multisensor
data fusion
”.
Instrume
ntatio
n
&
Measurement M
agazine, I
E
EE
. 2
008; 11(3): 24-3
3
.
[55]
LC Bento
,
U Nunes, F Moita, an
d A Surrecio. “
S
e
nsor fusion for precise autonom
ous
vehicle na
vigation in ou
tdoo
r
semi-structured environments
”
.
i
n
Intell
igent
Tra
n
sportation S
y
st
em
s, 2005. Proceedings. 2005 I
EEE
. IEE
E
. 200
5:
245-250.
[56]
L Jetto, S Longhi, and D Vitali. “Localization o
f
a wheeled
mobile robot b
y
sensor data fusion based on a fuzzy
logic
adap
ted
ka
lm
an fil
t
er
”.
Control Engin
eering
Practice
. 1999
;
7(6): 763-771
.
[57]
HP Moravec. “Sensor fusion in
cert
ainty
gr
ids for
mobile robo
ts”.
AI magazine
. 19
88; 9(2): 61.
[58]
S Ahn, J Choi,
NL Doh,
and
WK Chung.
“A practical appro
ach for
ekf-sla
m in
an indoor
en
vironment: fusin
g
ultras
oni
c s
e
ns
or
s
and s
t
er
eo
cam
era”
.
Autonomou
s robots
. 2008; 2
4
(3): 315-335
.
[59]
JA Ma
lpic
a,
MC Alonso,
a
nd
MA Sanz. “Dempster-shafer th
eor
y
in g
e
ograp
hic information
s
y
stems: A survey
”.
Ex
pe
rt Sy
ste
m
s
with Applic
ations
. 2007; 32
(1): 4
7
-55.
[60]
IL Davis and A
Stentz. “
Sensor fusion for auto
nomous outdoor
navigation usin
g neural networks
”. in
Inte
llig
e
n
t
Robots and S
y
stem
s 95.'
H
um
an Robot Intera
ction
and Coo
p
erat
ive Robots
'
, Proc
eed
ings.
1995 IEEE/RSJ
International Co
nference on
. 199
5; 3: 338-343.
[61]
N Ghosh, Y Ravi, A Patra, S Mukhopadh
y
a
y
,
S P
a
ul, A Mohanty
,
and A Chattop
a
dh
y
a
y
.
“Estimation of tool wear
during CNC m
illing using neural
network-based
sensor fu
sion”. Mechani
cal S
y
s
t
em
s and Signal Processing. 2007;
21(1): 466-479
.
[62]
M Mucientes, D
L
Moreno, A Bugarin,
a
nd S Barro. “Design of a fuzzy
contro
ll
er in mobile robotics using genetic
algorithm
s
”.
Ap
plied
Soft Computing
. 2007; 7(2)
: 540-546.
[63]
B Moshiri, M Reza Asharif
,
and
R Hosein Nezhad. “
P
seudo
inform
ation m
easure
:
A new concep
t
for extension of
bay
e
sian
fusion
in robotic map bu
ilding”.
In
formation Fusion
. 2002
; 3(1): 51-68.
[64]
K Nagla, M Uddin,
D
Singh
,
and
R Kumar. “
Obj
ect
iden
tif
icat
ion
in d
y
namic
environment using s
e
nsor fusion
”.
i
n
IEEE
39th
Workshop on Applied
Im
ager
y
Pattern
R
ecogn
ition
Workshop (AIPR), 2010. IEE
E
. 20
10: 1-4.
[65]
DE Rum
e
lhart
,
GE Hinton,
and
RJ W
illiam
s
. “
L
earn
i
ng repr
esenta
tions b
y
ba
c
k
-propagat
i
ng er
rors”.
Cognitive
modeling
. 2002;
1: 213.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
089
-48
56
IJR
A
V
o
l
.
3,
No
. 2,
J
u
ne 2
0
14:
1
3
1
– 13
8
13
8
[66]
JW van Dam,
BJ Krose, and FC Gr
oen. “Ad
a
ptiv
e sensor models”.
in Multisensor Fusion a
nd Integration for
Intelligen
t Systems, 1996. I
E
EE/SICE/
RS
J International Con
f
eren
ce on
. I
E
EE
. 199
6: 705-712.
[67]
M Kam,
X Zhu, and P
Kalata. “
Sensor fusion fo
r mobile robot n
a
vigation
”. Proceedings of the IEEE. 1997; 85(1):
108-119.
[68]
YF Zheng. “Integration of multiple sensors into a robo
tic s
y
stem and its performan
ce evaluatio
n
”. Robotics an
d
Automation, I
E
EE
Transactions
on. 1989; 5(5)
:
658-669.
[69]
N Yadaiah, L Singh, RS Bapi,
VS Rao,
BL Deekshatu
l
u, and
A Negi. “Mul
t
i
sensor data fusion using neural
networks”.
in N
e
ural Networks, 2
006. IJCNN'06
.
In
ternational Jo
int C
onference o
n
. IEEE
. 2006
: 8
75-881.
[70]
J Borenstein
an
d Y Koren. “Histogram
ic in-mo
tion mapping
fo
r mobile r
obo
t o
b
stacle avo
i
dan
c
e”. Robotics an
d
Automation, I
E
EE
Transactions
on. 1991; 7(4)
:
535-539.
[71]
M Dekhil and TC Henderson. “
I
n
strume
nte
d
logic
a
l se
nsor systems-practice
”. in Robotics and Automation, 1998
.
Proceedings. 19
98 IEE
E
In
terna
t
ional
Confere
n
c
e
on. 1998;
4:
31
03-3108.
[72]
H Alex,
M Kumar,
and B Shirazi.
“Midfu
sion: An adaptive midd
leware fo
r in
for
m
ation fusion in
sensor network
applications”.
In
formation Fusio
n
. 2008; 9(3): 33
2-343.
[73]
J Gonzale
z
,
JL
Blanco
, C Ga
lin
do, A Orti
z-de
Galisteo
,
J
A
F
e
r
n
andez-M
a
dr
iga
l
, F
A
Moreno, and JL Martinez.
“Mobile robot lo
calization b
a
sed
on ultra-wid
e
-b
and ranging: A particle f
ilter appro
ach”.
Robo
tics and autonomou
s
sy
ste
m
s
. 2009; 5
7
(5): 496-507
.
[74]
F Ferreira, I
Amorim, R Roch
a,
and
J Dias.
“T-slam: Registering
topologi
cal and
geometr
i
c maps for robo
t
localization
in l
a
rge environm
ent
s
”. in Mult
isensor Fu
sion and Integration for In
telligen
t S
y
st
em
s, 2008. MFI 2008
.
IEEE Internation
a
l Conf
eren
ce o
n
. IEEE. 2008
: 3
92-398.
BIOGRAP
HI
ES OF
AUTH
ORS
KS
Nagla is
P
h
.D. r
e
s
earch
s
c
holar
in
the
Department of
Instru
mentatio
n and Control
Engineering, at
Dr. B.R. Am
bedkar Nat
i
onal
Institute of T
echnolog
y
Jalandhar. He h
a
s
com
p
leted his
B.Te
ch. (Degre
e) in elec
troni
c
and ins
t
rum
e
ntation from
P
u
njabi Univers
i
t
y
Patial
a
and his
M.Tech in
instru
mentation and
control engin
eer
i
ng (special
iz
atio
n robotics) from
Dr. B.R. Am
bedkar Nation
a
l I
n
stitute of
Tec
hnolog
y
Jaland
har. He, with
his team
, has
develop
e
d sever
a
l robots
and new mechanisms during
the p
a
st ten
y
e
ars. He is
an inven
t
or of
three inv
e
ntions
(Indian patents)
and man
y
more
inventions ar
e in
the process of being granted.
His current
ar
ea
or research
is ar
t
i
ci
al
intellig
ence in m
obil
e
robo
ts and
industri
a
l
autom
a
tion.
Dr. Moin Uddin
completed his
Ph.D. from the
I
ndian Institute
of Technolog
y
(
IIT) Roorkee,
India in
y
e
ar 1
993. His area
of research
is
robotics,
computer network
i
ng, AI and Soft
Computing. He is a member
of various n
a
ti
onal and
in
tern
ation
a
l techn
i
c
a
l
professiona
l
bodies/societies. He is recip
i
ent
of the Dr
. Radh
a Krishnan Memorial Award-96
.
At present h
e
is
working as
P
r
o-
Vice-Chan
cel
lor
at
Delh
i T
echno
logic
a
l Univ
ers
i
t
y
D
e
lhi-Ind
i
a
.
Dr. Dilbag Sin
gh received
the B.E. (Hons.)
de
gree
in
ele
c
tr
ica
l
engin
eer
ing
from
P
unjab
Engineering College, Chandig
a
r
h
in 1991, and the M.E. deg
r
ee
in
control and gu
idance from the
Universit
y
of R
oorkee in 1993,
and the Ph.D. degr
ee in engin
eer
ing from
the Ind
i
an Institut
e
of
Techno
log
y
Roo
r
kee, in 2004. H
e
is presently
se
r
v
ing as Associate Professor of In
strumentation
and Control
Eng
i
neer
ing at Dr
.
B.R. Am
bedkar
National Institu
t
e
of Technolog
y Jalandhar
.
His
res
earch
in
teres
t
s
are
in s
i
gn
al
pr
oces
s
i
ng, sensors and biomed
ical applications.
Evaluation Warning : The document was created with Spire.PDF for Python.