I
AE
S
I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
Ar
t
if
icial
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
Vol.
14
,
No.
4
,
Augus
t
2025
,
pp.
2805
~
2814
I
S
S
N:
2252
-
8938
,
DO
I
:
10
.
11591/i
jai
.
v
14
.i
4
.
pp
28
05
-
2814
2805
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
ai
.
iaes
c
or
e
.
c
om
A
n
au
t
om
at
i
c
soc
ia
l
e
n
gag
e
m
e
n
t
m
e
as
u
r
e
m
e
n
t
d
u
r
in
g
h
u
m
an
-
r
ob
ot
i
n
t
e
r
ac
t
io
n
Wae
l
Has
an
Ali
Al
m
oh
am
m
e
d
1
,
S
in
an
Adn
an
M
u
h
is
n
2
,
Z
ah
r
aa
Abed
Alj
as
im
M
u
h
is
n
3
1
D
e
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ie
nc
e
,
C
ol
le
g
e
of
C
omput
e
r
S
c
ie
nc
e
a
nd
I
nf
or
ma
ti
on
T
e
c
hnol
ogy
,
U
ni
ve
r
s
it
y of
K
e
r
ba
la
,
K
a
r
ba
la
,
I
r
a
q
2
C
ol
le
ge
of
B
io
te
c
hnol
ogy, Al
-
Q
a
s
im
G
r
e
e
n U
ni
ve
r
s
it
y,
B
a
byl
on,
I
r
a
q
3
C
omput
e
r
C
e
nt
e
r
, A
l
-
Q
a
s
im
G
r
e
e
n U
ni
ve
r
s
it
y, B
a
byl
on, I
r
a
q
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
Oc
t
7,
2024
R
e
vis
e
d
F
e
b
17,
2025
Ac
c
e
pted
M
a
r
15,
2025
So
ci
a
l
en
g
ag
eme
n
t
refers
t
h
e
ex
p
re
s
s
i
o
n
s
o
f
ex
i
s
t
i
n
g
i
n
t
erp
er
s
o
n
al
rel
at
i
o
n
s
h
i
p
s
d
u
ri
n
g
t
h
e
i
n
t
era
ct
i
o
n
w
h
i
c
h
rep
re
s
en
t
s
t
h
e
act
u
al
i
n
t
ere
s
t
i
n
g
o
f
h
u
ma
n
i
n
t
h
e
i
n
t
erac
t
i
o
n
.
H
o
w
ev
er,
s
o
ci
a
l
en
g
a
g
eme
n
t
meas
u
reme
n
t
i
s
a
s
i
g
n
i
f
i
ca
n
t
co
n
cer
n
i
n
s
o
c
i
al
h
u
ma
n
-
ro
b
o
t
i
n
t
erac
t
i
o
n
(
H
RI)
b
ecau
s
e
o
f
i
t
s
ro
l
e
i
n
u
n
d
ers
t
an
d
i
n
g
t
h
e
i
n
t
erac
t
i
o
n
’
s
t
ren
d
an
d
a
d
a
p
t
ro
b
o
t
’s
b
e
h
av
i
o
r
acco
rd
i
n
g
l
y
.
H
en
ce,
w
e
ac
h
i
e
v
ed
t
h
e
t
w
o
mai
n
o
b
j
e
ct
i
v
es
o
f
t
h
i
s
s
t
u
d
y
.
F
i
rs
t
l
y
,
en
r
i
ch
me
n
t
t
h
e
t
h
e
o
ret
i
cal
l
i
t
era
t
u
re
an
d
rel
at
e
d
co
n
cep
t
s
.
Seco
n
d
l
y
,
p
ro
p
o
s
ed
a
ro
b
u
s
t
n
e
u
ra
l
n
et
w
o
r
k
mo
d
el
w
h
i
c
h
i
s
mu
l
t
i
l
ay
er
p
erce
p
t
r
o
n
(ML
P)
cl
as
s
i
f
i
er
t
o
meas
u
re
s
o
c
i
al
en
g
ag
eme
n
t
s
t
a
t
e
d
u
ri
n
g
i
n
t
e
ract
i
o
n
.
PIn
So
Ro
d
at
as
e
t
w
a
s
u
s
ed
f
o
r
t
ra
i
n
i
n
g
an
d
t
e
s
t
i
n
g
p
u
rp
o
s
e.
In
p
art
i
cu
l
ar,
t
h
e
p
aramet
er
s
o
f
ML
P
mo
d
el
w
ere
met
i
cu
l
o
u
s
l
y
craft
ed
t
o
reco
g
n
i
ze
t
h
e
s
o
ci
a
l
en
g
a
g
emen
t
accu
rat
e
l
y
.
W
e
ev
al
u
a
t
ed
t
h
e
mo
d
el
’s
p
er
fo
rman
ce
b
y
s
e
v
eral
met
ri
c
s
an
d
t
h
e
res
u
l
t
s
h
o
w
ed
a
n
i
n
t
eres
t
i
n
g
acc
u
ra
cy
reach
ed
9
4
.
8
5
%
.
G
i
v
en
t
h
a
t
,
i
t
s
u
p
p
o
rt
s
t
h
e
ro
b
o
t
t
o
h
as
ad
ap
t
i
v
e
an
d
res
p
o
n
s
i
v
e
b
e
h
av
i
o
r
in
real
t
i
me
a
p
p
l
i
ca
t
i
o
n
s
w
h
i
ch
i
s
i
m
p
ro
v
i
n
g
H
RI
ev
en
t
u
a
l
l
y
.
K
e
y
w
o
r
d
s
:
Huma
n
-
r
obot
int
e
r
a
c
ti
on
Ne
ur
a
l
ne
twor
k
S
oc
ial
e
nga
ge
ment
S
oc
ial
r
oboti
c
s
Us
e
r
e
nga
ge
ment
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
W
a
e
l
Ha
s
a
n
Ali
Almohammed
De
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ienc
e
,
C
oll
e
ge
o
f
C
om
puter
S
c
ienc
e
a
nd
I
n
f
or
mation
T
e
c
hnology
Unive
r
s
it
y
of
Ke
r
ba
la
Ka
r
ba
la,
I
r
a
q
E
mail:
wa
e
l
.
h@uoke
r
ba
la
.
e
du.
iq
1.
I
NT
RODU
C
T
I
ON
R
e
c
e
ntl
y,
s
oc
ial
s
e
r
vice
s
r
obots
a
s
a
n
a
s
s
is
tant
or
a
c
ompanion
ha
ve
be
gun
int
e
gr
a
ted
to
our
s
e
r
vice
s
e
nvir
onments
.
T
he
y
a
r
e
pe
r
va
s
ively
tur
ning
int
o
pa
r
t
of
e
ve
r
yda
y
tas
ks
in
e
duc
a
ti
on,
wor
k
,
a
nd
he
a
lt
hc
a
r
e
.
Huma
n
-
r
obot
int
e
r
a
c
ti
on
(
HR
I
)
a
nd
s
oc
ial
r
ob
oti
c
s
s
tudy
how
r
obots
s
uppor
t
human
thr
oug
h
s
oc
ial
int
e
r
a
c
ti
on
with
a
n
ins
ight
on
de
ve
lopi
ng
a
n
in
t
e
r
a
c
ti
on
with
indi
viduals
in
di
f
f
e
r
e
nt
c
ontexts
e
f
f
e
c
ti
ve
ly
[
1]
,
[
2]
.
Ge
ne
r
a
ll
y,
s
oc
ial
r
obots
a
r
e
de
s
igned
a
s
us
e
r
-
f
r
iendly
e
ve
n
f
or
us
e
r
s
without
tec
hnologi
c
a
l
ba
c
kgr
ound
s
uc
h
a
s
c
hil
dr
e
n.
R
e
s
e
a
r
c
he
s
in
t
he
s
e
f
ields
ha
ve
f
oc
us
e
d
on
the
f
a
c
tor
s
that
i
nf
luenc
e
indi
viduals
’
be
ha
vior
a
nd
pe
r
c
e
pti
on
towa
r
d
r
obot
s
[
3]
.
De
f
ini
tely,
c
hil
d
-
r
obot
int
e
r
a
c
ti
on
is
a
n
e
s
s
e
nti
a
l
a
nd
c
r
it
ica
l
r
e
s
e
a
r
c
h
f
ield
a
s
s
oc
ial
r
obo
ts
a
r
e
s
igni
f
ica
ntl
y
e
mpl
oye
d
to
wor
k
with
.
C
hil
d
r
e
n
a
r
e
int
e
r
a
c
t
ing
with
r
obots
in
dif
f
e
r
e
nt
wa
y
s
ince
they
ha
ve
dif
f
e
r
e
nt
i
mm
a
tur
e
c
ognit
ive
de
ve
lopm
e
nt
a
nd
da
il
y
l
ivi
ng
s
kil
ls
a
s
we
ll
a
s
they
ha
ve
high
a
bil
it
y
to
a
da
pt
a
nd
lea
r
n
n
e
w
tec
hnology
[
3
]
,
[
4
]
.
Nor
mally
,
c
hil
dr
e
n
do
no
t
int
e
r
a
c
t
with
r
obot
a
s
a
mec
ha
tr
onic
de
vice
with
a
c
omput
e
r
pr
og
r
a
m,
bu
t
the
c
ha
r
a
c
ter
is
ti
c
s
of
r
obot
thes
e
a
r
e
us
ua
ll
y
e
xpe
c
ted
to
be
s
im
il
a
r
to
a
ny
li
v
ing
s
ys
tem.
F
ur
th
e
r
mor
e
,
the
pe
r
s
pe
c
ti
ve
s
of
c
hil
d
r
e
n
towa
r
d
r
o
bot
s
a
r
e
f
a
r
dif
f
e
r
e
nt
f
r
om
thos
e
of
a
dult
s
.
He
nc
e
,
e
xpa
nding
t
his
knowle
dge
to
c
hil
dr
e
n’
s
be
ha
vior
is
c
r
uc
ial
to
pos
it
ively
e
nga
ge
with
r
obot.
Ge
ne
r
a
ll
y
,
r
obot
’
s
a
tt
it
ude
e
f
f
e
c
ted
dir
e
c
tl
y
on
e
nga
ge
ment
of
c
hil
d
with
th
e
r
obot.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
4,
Augus
t
2025
:
280
5
-
2814
2806
P
r
oduc
ti
vit
y
a
nd
qua
li
ty
o
f
int
e
r
a
c
ti
on
a
r
e
va
s
tl
y
c
or
r
e
late
d
with
incr
e
a
s
ing
of
e
nga
ge
ment
leve
l
[
5]
,
[
6]
.
T
he
r
e
f
or
e
,
we
us
e
c
oll
is
ion
r
is
k
index
(
CRI
)
a
s
a
c
a
s
e
to
be
s
tudi
e
d
in
thi
s
pa
pe
r
.
E
nga
ge
ment
c
onc
e
pt
is
br
oa
dly
s
tudi
e
d
in
HR
I
a
s
a
c
or
e
is
s
ue
in
the
int
e
r
a
c
ti
on.
Although,
the
mea
ning
of
thi
s
te
r
m
doe
s
not
ha
ve
a
n
e
xpli
c
it
de
f
ini
ti
on
ye
t
.
Ge
ne
r
a
ll
y
,
it
r
e
f
e
r
s
to
be
ing
invol
ve
d
i
n
f
or
mal
or
inf
o
r
mal
s
oc
ial
a
c
ti
vit
ies
.
How
e
ve
r
,
s
ome
r
e
s
e
a
r
c
he
r
s
ha
ve
de
f
ined
it
a
s
“
the
pr
oc
e
s
s
by
whic
h
in
ter
a
c
tor
s
s
tar
t,
maintain
a
nd
e
nd
their
pe
r
c
e
ived
c
onne
c
ti
on
to
e
a
c
h
other
dur
ing
a
n
int
e
r
a
c
ti
on”
[
7
]
,
[
8]
.
T
ypica
ll
y,
e
nga
ge
ment
leve
l
e
xpr
e
s
s
e
s
how
the
int
e
r
a
c
ti
on
be
twe
e
n
human
a
nd
r
obot
is
s
uc
c
e
s
s
f
ul.
I
nde
e
d,
the
ke
y
goa
l
is
to
s
us
tain
the
human
e
nga
ge
d
du
r
ing
the
int
e
r
a
c
ti
on.
F
ur
ther
mor
e
,
e
nga
ge
ment
leve
l
c
a
n
inf
lu
e
nc
e
the
int
e
r
a
c
ti
on
s
tr
a
tegy
whe
r
e
a
s
if
the
f
luctua
ti
on
in
us
e
r
e
nga
ge
ment
is
a
ble
to
be
de
tec
ted,
the
int
e
r
a
c
ti
on
s
tr
a
tegy
c
ould
be
f
or
mul
a
ted
to
e
nf
or
c
e
the
us
e
r
s
e
xpe
r
ienc
e
a
nd
ke
e
p
them
e
nga
ge
d.
I
n
a
ddit
ion,
r
e
a
li
z
ing
us
e
r
’
s
e
nga
ge
ment
is
s
igni
f
ica
nt
to
pr
ov
ide
pe
r
s
ona
li
z
e
d
s
uppor
t
a
nd
a
void
d
r
opouts
.
T
he
r
e
f
or
e
,
mea
s
ur
e
ment
of
e
nga
ge
ment’
s
leve
l
is
a
pivot
a
l
f
unc
ti
on
in
HR
I
[
9]
–
[
11]
.
Ac
c
or
dingl
y
,
tr
a
c
ing
human’
s
e
nga
ge
ment
ha
s
be
e
n
a
pr
omi
s
ing
r
e
s
e
a
r
c
h
a
r
e
a
.
How
e
ve
r
,
ther
e
a
r
e
two
main
methods
to
mea
s
ur
e
the
e
nga
ge
ment
w
hich
a
r
e
a
utom
a
ti
c
a
nd
manua
l
[
12]
.
I
n
tr
a
dit
ional
wa
y,
a
thi
r
d
pa
r
ty
c
a
n
r
e
c
ognize
the
e
nga
ge
ment
leve
l
by
dir
e
c
t
obs
e
r
va
ti
on
c
he
c
kli
s
t
a
nd
r
a
te
s
c
a
le.
On
the
other
ha
nd,
us
ing
a
lea
r
ning
s
ys
tem
f
or
a
utom
a
ti
c
e
ng
a
ge
ment
mea
s
ur
e
ment
[
9]
.
T
he
r
e
a
r
e
s
e
ve
r
a
l
f
or
ms
of
e
nga
ge
ment
s
uc
h
a
s
a
f
f
e
c
ti
ve
e
nga
ge
ment,
s
oc
ial
e
nga
ge
ment,
a
nd
c
ognit
ive
e
nga
ge
ment.
S
ince
the
c
onc
e
pt
o
f
e
nga
ge
ment
it
s
e
lf
is
ye
t
unc
lea
r
,
thus
ther
e
is
no
plain
e
xplan
a
ti
on
f
or
e
a
c
h
f
or
m
with
it
s
f
e
a
tur
e
s
a
nd
ther
e
is
a
n
ove
r
lapping
in
the
de
f
ini
ti
on
of
e
a
c
h
f
or
m
.
How
e
ve
r
,
mos
t
of
e
xis
ti
ng
wor
k
c
onc
e
ntr
a
ted
on
c
ognit
ive
a
nd
e
mot
ional
e
nga
ge
ment
s
ince
thes
e
f
or
ms
a
r
e
mor
e
de
f
ined
a
nd
unde
r
s
tood
to
s
ome
e
xtent
s
o
th
e
y
a
r
e
e
a
s
ier
to
be
r
e
c
ognize
d
while
the
s
oc
ial
e
ng
a
ge
ment
f
or
m
ha
s
gott
e
n
les
s
a
tt
e
nti
on
[
11]
,
[
13]
.
On
other
ha
nd,
by
pe
r
us
ing
the
li
te
r
a
tur
e
,
r
e
s
e
a
r
c
he
r
s
us
e
d
diver
s
e
methods
to
mea
s
ur
e
the
s
tate
of
e
nga
ge
ment
f
or
ms
.
M
a
c
hine
lea
r
ning
a
nd
de
e
p
lee
r
ing
models
h
a
ve
be
e
n
e
mpl
oye
d
in
the
mos
t
due
to
the
f
a
c
t
that
the
y
ha
ve
be
e
n
pr
ove
n
their
e
f
f
icie
nc
y
in
f
ield
of
pa
tt
e
r
n
r
e
c
ognit
ion
e
s
pe
c
ially
with
the
quick
a
nd
mas
s
ive
a
dva
nc
e
ment
in
the
c
omput
a
ti
ona
l
s
of
twa
r
e
a
nd
ha
r
dwa
r
e
[
14]
–
[
16]
.
Ne
ve
r
thele
s
s
,
the
va
s
t
major
it
y
of
the
p
r
e
vious
s
tudi
e
s
ha
ve
be
e
n
mea
s
ur
ing
the
e
nga
ge
m
e
nt
s
tate
,
r
e
ga
r
dles
s
it
s
f
or
m,
by
us
ing
binar
y
c
las
s
if
ica
ti
on
of
two
c
las
s
e
s
whic
h
a
r
e
e
nga
ge
d
or
not
e
nga
ge
d
while
ther
e
a
r
e
a
r
a
nge
of
e
nga
ge
ment
s
tate
s
in
be
twe
e
n,
e
a
c
h
one
c
ould
be
im
p
r
ove
d
dif
f
e
r
e
ntl
y.
T
he
r
e
f
or
e
,
thi
s
pa
pe
r
indi
c
a
tes
two
r
e
s
e
a
r
c
h
que
s
ti
ons
to
be
a
ns
we
r
e
d
:
wha
t
is
the
de
f
ini
ti
on
o
f
e
nga
ge
ment
in
HR
I
a
nd
it
s
c
omponents
’
c
ha
r
a
c
ter
is
ti
c
s
a
nd
how
to
de
ve
lop
a
n
im
p
r
ove
d
a
utom
a
ti
c
e
nga
ge
ment
mea
s
ur
e
ment
model
c
ompar
ing
to
e
xis
ti
ng
s
tudi
e
s
f
oc
us
ing
on
s
oc
ial
e
nga
ge
ment
pa
r
t
icula
r
ly
.
I
n
or
de
r
to
a
ns
we
r
thes
e
que
s
ti
ons
,
we
s
e
t
two
o
bjec
ti
ve
s
f
or
thi
s
s
tudy
whic
h
a
r
e
to
im
pl
icitl
y
d
e
f
ine
the
c
onc
e
pt
of
e
nga
ge
ment
a
nd
und
e
r
s
tand
the
c
ha
r
a
c
ter
is
ti
c
s
of
e
a
c
h
f
or
m.
As
we
ll
a
s
,
e
xtend
the
r
e
s
e
a
r
c
h
by
de
ve
lopi
ng
a
n
e
f
f
icie
nt
model
to
a
utom
a
ti
c
a
ll
y
mea
s
ur
e
the
s
oc
ial
e
ng
a
ge
ment
f
or
m
s
pe
c
if
ica
ll
y.
T
his
wor
k
pr
opos
e
d
a
ne
ur
a
l
ne
twor
k
model
to
mea
s
ur
e
s
oc
ial
e
nga
ge
ment
leve
l
dur
ing
C
R
I
.
T
his
wor
k
is
ti
ghtl
y
r
e
leva
nt
to
the
a
utom
a
ti
c
us
e
r
a
c
ti
vit
ies
r
e
c
ognit
ion.
I
t
ha
s
us
e
d
mul
ti
modal
da
tas
e
t
c
ompos
ing
of
vis
ua
l
a
nd
a
udio
modalit
ies
.
Additi
ona
ll
y
,
the
pr
opos
e
d
mode
l
c
las
s
if
ies
e
nga
ge
ment
s
tate
int
o
mul
ti
ple
c
las
s
e
s
.
T
he
r
e
s
t
of
pa
pe
r
is
s
tr
uc
tu
r
e
d
a
s
f
oll
ows
:
s
e
c
ti
on
2
dis
c
us
s
e
s
the
de
f
ini
ti
ons
o
f
e
nga
ge
ment
c
onc
e
pt,
types
,
mea
s
ur
e
ment
a
pp
r
oa
c
he
s
,
a
nd
a
s
na
ps
hot
of
r
e
late
d
wo
r
k.
T
he
method
de
tails
including
r
e
s
e
a
r
c
h
de
s
ign,
da
tas
e
t
a
nd
a
na
lys
is
pr
oc
e
s
s
,
a
nd
e
xpe
r
i
menta
l
e
xa
mi
ne
s
s
ys
te
matica
ll
y
in
s
e
c
ti
on
3.
F
i
na
ll
y,
the
r
e
s
ult
,
c
onc
lus
ion
a
nd
f
u
tur
e
di
r
e
c
ti
ons
e
va
luate
s
a
nd
dis
c
us
s
e
s
in
s
e
c
ti
on
s
4
a
nd
5
r
e
s
pe
c
ti
ve
ly.
2.
B
AC
KG
ROUN
D
AN
D
RE
L
AT
E
D
S
T
UD
I
E
S
T
his
s
e
c
ti
on
c
ove
r
s
the
main
c
onc
e
pt
s
a
nd
c
onte
xt
whic
h
is
ne
c
e
s
s
a
r
y
to
unde
r
s
tand
the
r
e
s
e
a
r
c
h
pr
oblem.
I
t
be
gins
with
theor
e
ti
c
a
l
ba
c
kgr
oun
d
s
uc
h
a
s
the
e
nga
ge
ment
c
onc
e
pt
in
HR
I
,
i
t
f
o
r
ms
,
mea
s
ur
e
ment
a
ppr
oa
c
h.
E
ve
ntually,
it
high
li
ghts
t
he
r
e
late
d
s
tudi
e
s
,
identif
ying
the
main
methods
,
f
indi
ngs
,
a
nd
e
xis
ti
ng
pr
oblems
:
2.
1
.
E
n
gage
m
e
n
t
in
h
u
m
a
n
-
r
ob
ot
in
t
e
r
ac
t
ion
I
n
the
de
ve
lopm
e
nt
of
s
oc
ial
int
e
ll
igent
tec
hnolog
y
s
uc
h
a
s
(
r
obot,
c
omput
e
r
,
o
r
vir
tual
a
ge
nt)
,
the
r
e
a
r
e
dif
f
e
r
e
nt
is
s
ue
s
s
ha
ll
be
c
ons
ider
e
d
in
or
de
r
to
pe
r
s
ona
li
z
e
the
int
e
r
a
c
ti
on.
I
nde
e
d,
e
nga
ge
ment
i
s
one
of
m
a
in
thes
e
is
s
ue
s
that
br
oa
dly
uti
li
z
e
d
a
s
a
ke
y
s
oc
ial
phe
nomenon
in
the
HR
I
f
ield
[
17]
.
T
he
r
e
s
e
a
r
c
h
f
il
e
d
of
e
nga
ge
ment
r
obots
wi
th
pe
ople
(
us
e
r
s
)
is
obtai
ning
a
n
int
e
ns
ive
a
tt
e
nti
on
a
nd
int
e
r
e
s
t
a
mong
r
e
s
e
a
r
c
he
r
s
[
18]
.
R
e
ga
r
dles
s
the
c
omm
on
us
e
o
f
e
nga
ge
men
t,
the
r
e
is
no
e
xpli
c
it
mea
ning
or
int
e
r
p
r
e
tation
c
onc
e
pt.
C
onve
r
s
e
ly,
the
de
f
ini
t
ion
o
f
e
nga
ge
ment
is
ye
t
c
ha
r
a
c
ter
ize
d
by
a
mbi
guit
y
a
nd
big
va
r
iation
[
1
9]
,
[
20]
.
How
e
ve
r
,
s
ome
s
tudi
e
s
de
f
ine
the
c
onc
e
pt
of
e
n
ga
ge
ment
with
the
tec
hnologi
e
s
a
nd
it
s
r
ole
in
p
a
r
ti
c
ular
c
ontexts
.
T
o
de
mons
tr
a
te,
S
idne
r
e
t
al
.
[
7]
wa
s
f
r
om
the
e
a
r
li
e
r
to
de
f
ine
e
nga
ge
ment
c
onc
e
pt,
in
ge
ne
r
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
A
n
automati
c
s
oc
ial
e
ngage
me
nt
me
as
ur
e
me
nt
dur
ing
human
-
r
obot
…
(
W
ae
l
Has
an
A
li
A
lmohamm
e
d
)
2807
f
or
m,
a
s
“
the
pr
oc
e
s
s
by
whic
h
two
(
or
mo
r
e
)
pa
r
ti
c
ipants
e
s
tablis
h,
maintain
a
nd
e
nd
thei
r
pe
r
c
e
ived
c
onne
c
ti
on
dur
ing
int
e
r
a
c
ti
ons
they
joi
ntl
y
unde
r
ta
ke
”
.
L
a
ter
o
n,
P
ogg
i
[
21
]
de
f
i
ne
b
y
us
ing
de
e
pe
r
ter
m
s
a
s
“
the
va
l
ue
that
a
pa
r
t
icip
a
nt
in
a
n
int
e
r
a
c
ti
on
a
tt
r
ibu
tes
t
o
the
goa
l
of
be
ing
to
ge
the
r
wi
th
the
ot
he
r
pa
r
t
icipa
nt(
s
)
a
nd
c
on
ti
n
uing
int
e
r
a
c
t
ion”
.
I
n
HR
I
c
ontex
t,
e
n
ga
ge
me
nt
is
a
c
onc
e
p
t
of
t
he
gr
e
a
tes
t
s
ign
if
ica
nc
e
du
e
to
it
s
a
b
il
i
ty
of
s
ha
ping
the
de
s
ign
of
,
de
ve
lop
ing
a
m
or
e
a
d
va
nc
e
d
,
a
n
d
a
da
pta
ble
i
nte
r
f
a
c
e
s
f
or
us
e
r
s
a
s
we
ll
a
s
c
ont
r
i
but
ion
t
o
be
tt
e
r
i
nt
e
r
a
c
ti
on
outco
me.
How
e
ve
r
,
e
nga
ge
m
e
n
t
ha
s
a
d
yna
m
ic
n
a
tur
e
whic
h
mea
ns
i
t
is
c
ha
ngi
ng
ove
r
t
im
e
a
nd
be
twe
e
n
int
e
r
a
c
ti
ons
.
W
it
h
the
na
tu
r
e
of
e
nga
ge
ment
in
m
i
nd
a
nd
r
e
f
e
r
r
in
g
to
de
f
ini
ti
o
n
o
f
P
o
ggi
[
2
1]
,
e
n
ga
g
e
ment
is
c
ons
ide
r
e
d
a
qua
l
it
y
mea
s
u
r
e
o
f
the
int
e
r
a
c
t
ion
.
C
ons
ide
r
ing
tha
t,
O'B
r
ien
a
nd
T
o
ms
de
f
ine
d
t
he
e
n
ga
ge
men
t
a
s
“
a
q
ua
l
it
y
of
us
e
r
e
x
pe
r
ienc
e
c
ha
r
a
c
ter
ize
d
b
y
a
t
tr
i
butes
o
f
c
h
a
ll
e
nge
,
pos
i
ti
ve
a
f
f
e
c
t
,
e
n
du
r
a
bi
li
t
y,
a
e
s
thetic
a
nd
s
e
ns
o
r
y
a
p
pe
a
l
,
a
tt
e
nti
on
,
f
e
e
d
ba
c
k
,
va
r
iety
/no
ve
lt
y
,
in
ter
a
c
ti
v
it
y
,
a
n
d
pe
r
c
e
ived
us
e
r
c
ont
r
o
l”
[
22
]
.
De
f
ini
tely
,
the
ult
im
a
te
goa
l
o
f
HR
I
is
to
e
s
tablis
h
a
high
leve
l
of
e
nga
ge
ment
dur
ing
int
e
r
a
c
ti
on,
c
ons
e
que
ntl
y,
a
c
hieve
the
int
e
r
a
c
ti
on’
s
tas
k
s
uc
c
e
s
s
f
ull
y.
He
nc
e
,
r
e
inf
o
r
c
e
ment
of
e
nga
ge
ment
e
nha
nc
e
s
the
qua
li
ty
of
int
e
r
a
c
ti
on
whic
h
r
e
f
lec
ted
e
ve
ntually
o
n
incr
e
a
s
ing
the
pos
s
ibi
li
ty
o
f
a
c
hieving
int
e
r
a
c
ti
o
n’
s
goa
l
[
19]
,
[
20]
.
S
o
that
,
mea
s
ur
ing
us
e
r
’
s
e
nga
ge
ment
c
a
n
give
ins
ight
f
or
de
ve
lopi
ng
the
us
e
r
int
e
r
a
c
ti
on
whe
r
e
a
s
li
ter
a
tur
e
s
a
mpl
y
c
onc
luded
the
pos
it
ive
r
e
lations
hip
be
twe
e
n
us
e
r
e
nga
ge
ment
a
nd
tas
k
a
c
hiev
e
ment.
R
obots
may
f
or
mul
a
te
int
e
r
a
c
ti
on
s
tr
a
tegy
to
s
us
tain
the
u
s
e
r
s
e
nga
ge
d
or
im
pr
ove
the
e
nga
ge
ment
leve
l,
i
f
they
got
the
a
bil
it
y
to
mea
s
ur
e
the
s
tate
o
f
us
e
r
e
nga
ge
ment
dur
ing
int
e
r
a
c
ti
on.
An
a
c
c
ur
a
te
e
nga
ge
ment
mea
s
ur
e
ment
c
a
n
s
uppor
t
r
obots
to
a
da
p
t
thei
r
be
ha
vior
in
or
de
r
to
inc
r
e
a
s
e
the
s
uc
c
e
s
s
of
int
e
r
a
c
ti
on’
s
tas
k
a
nd
e
nha
nc
e
us
e
r
e
xpe
r
ienc
e
[
23]
,
[
24
]
.
2
.
1.
1
.
E
n
gage
m
e
n
t
c
om
p
on
e
n
t
s
A
l
o
n
g
w
i
t
h
t
h
e
d
i
f
f
i
c
ul
t
y
o
f
s
ta
t
i
ng
a
n
e
x
pl
i
c
i
t
a
nd
c
o
m
p
r
e
he
ns
i
v
e
d
e
fi
n
i
t
i
on
o
f
e
n
g
a
g
e
me
n
t
te
r
m
,
m
a
n
y
s
t
u
d
ies
h
a
v
e
be
e
n
c
on
f
i
r
m
e
d
t
h
e
p
o
i
n
t
o
f
v
ie
w
t
he
e
n
g
a
g
e
me
n
t
is
a
c
o
mp
l
i
c
a
t
e
d
c
on
c
e
p
t
a
n
d
f
o
r
ms
o
f
m
u
l
t
i
p
le
c
o
m
p
on
e
n
ts
w
h
ic
h
a
r
e
r
e
l
e
va
n
t
a
m
on
g
t
h
e
ms
e
l
v
e
s
t
i
g
h
t
l
y
b
ut
t
h
e
y
a
r
e
s
t
i
l
l
de
t
e
c
t
e
d
by
p
a
r
t
i
c
u
l
a
r
i
n
d
ic
a
t
o
r
f
o
r
e
a
c
h
b
e
h
a
v
io
r
i
n
de
p
e
n
d
e
n
t
l
y
.
Ac
c
o
r
d
i
n
g
ly
,
e
ng
a
ge
m
e
n
t
is
d
iv
i
de
d
i
n
to
d
i
f
f
e
r
e
n
t
c
o
m
po
n
e
nt
s
o
f
e
ng
a
ge
m
e
n
t
b
y
d
i
f
f
e
r
e
n
t
w
o
r
k
s
u
c
h
a
s
c
o
gn
i
t
i
ve
,
a
f
f
e
c
t
i
v
e
,
be
h
a
v
i
or
a
l
,
s
o
c
i
a
l
,
a
n
d
t
a
s
k
.
A
ls
o
,
s
o
m
e
s
t
ud
i
e
s
c
on
s
i
d
e
r
e
d
a
h
y
b
r
i
d
e
n
g
a
g
e
me
n
t
c
o
m
p
o
ne
n
t
l
ik
e
s
o
c
ia
l
-
e
mo
t
i
o
na
l
,
s
oc
i
al
-
ta
s
k
,
a
n
d
s
o
c
ia
l
-
c
o
g
n
i
t
iv
e
[
9
]
,
[
2
4
]
–
[
2
8
]
.
I
n
t
h
is
s
tu
d
y
,
w
e
d
i
s
c
us
s
e
d
a
l
l
kn
o
w
n
in
d
i
v
id
u
a
l
c
o
m
po
n
e
n
t
s
a
s
f
o
l
low
s
.
a)
C
ognit
ive
e
nga
ge
ment
T
his
c
ompone
nt
of
e
nga
ge
ment
ha
s
be
e
n
typi
c
a
ll
y
invol
ve
d
c
ons
c
ious
c
omponents
li
ke
inves
tm
e
nt,
a
tt
e
nti
on,
a
nd
e
f
f
o
r
t
f
or
ins
tanc
e
whe
n
us
e
r
s
inves
t
their
c
ognit
ive
r
e
s
our
c
e
s
dur
ing
the
int
e
r
a
c
ti
on
a
w
a
y
f
r
om
e
mot
ional,
phys
ica
l,
or
s
oc
ial
r
e
s
our
c
e
s
to
r
e
inf
or
c
e
the
r
ol
e
of
pe
r
f
o
r
manc
e
(
e
.
g
.
I
ha
ve
to
wo
r
k
ha
r
d)
[
27]
,
[
29]
.
On
the
whole
,
c
ognit
ive
e
nga
ge
ment
c
onc
e
r
ns
of
how
the
us
e
r
s
buil
d
their
c
onne
c
ti
o
n
dur
ing
int
e
r
a
c
ti
on,
thi
nk
ing
a
c
ti
ve
ly,
a
ns
we
r
ing
the
que
s
ti
ons
,
a
nd
r
e
s
olvi
ng
the
pr
oblems
[
30
]
.
I
t
c
a
n
be
d
e
f
ined
a
s
the
e
f
f
or
ts
to
unde
r
s
tand
a
nd
a
na
lyze
the
int
e
r
a
c
ti
o
n
c
onc
e
pt
including
meta
-
c
ognit
ive
be
ha
vior
s
s
uc
h
a
s
how
the
us
e
r
s
e
t’
s
goa
l,
plans
,
a
nd
or
ga
nize
their
e
f
f
or
t
to
a
c
hieve
the
tas
k.
I
t
wa
s
a
ls
o
de
f
ined
a
s
a
n
int
e
ns
it
y
of
e
ngr
os
s
ment,
c
onc
e
ntr
a
ti
on,
a
nd
f
oc
us
to
a
c
hieve
t
he
tas
k
dur
ing
int
e
r
a
c
ti
on
[
31
]
,
[
32]
.
b)
B
e
ha
vior
a
l
e
nga
ge
ment
Ge
ne
r
a
ll
y,
it
r
e
f
e
r
s
to
us
e
r
a
tt
e
nti
on
towa
r
ds
tas
ks
c
ompl
e
ti
on
dur
ing
the
int
e
r
a
c
ti
on.
B
e
ha
vior
a
l
e
nga
ge
ment
ha
s
be
e
n
de
f
ined
a
s
a
pr
oa
c
ti
ve
pr
e
dis
pos
it
ion
of
u
s
e
r
to
a
dopt
with
the
c
ha
nge
s
a
nd
e
xpe
r
ienc
e
s
dur
ing
the
int
e
r
a
c
ti
on,
in
a
ddit
ion
,
the
de
s
ir
e
to
be
e
nha
nc
e
d
towa
r
d
thes
e
c
ha
nge
s
.
I
t
is
c
ons
ider
e
d
the
e
nc
our
a
ge
ment
that
mot
ives
the
pa
r
ti
c
ipation
in
th
e
tas
k
[
33]
.
B
e
ha
vior
a
l
e
nga
ge
ment
is
a
ddr
e
s
s
e
d
a
t
the
tas
k
leve
l
whe
n
ther
e
a
r
e
a
goa
l
-
or
iente
d
tas
ks
f
or
e
s
tablis
hing
the
e
nga
ge
ment.
T
he
r
e
f
o
r
e
,
a
s
long
a
s
be
ha
vio
r
a
l
e
nga
ge
ment
incr
e
a
s
e
s
,
the
mor
e
pos
it
ive
im
pa
c
t
it
ha
s
on
tas
k
a
c
hieve
ment
[
34]
.
T
h
is
c
omponent
of
e
nga
ge
ment
ha
s
be
e
n
f
ound
in
na
tur
e
,
pur
pos
e
,
la
c
k
of
di
f
f
iculty,
a
nd
f
a
mi
li
a
r
it
y
of
the
tas
k,
while
it
mi
s
s
e
s
e
mot
ional
a
nd
s
oc
ial
f
a
c
tor
s
.
T
he
ke
y
f
e
a
tur
e
o
f
thi
s
type
is
that
the
human
c
a
n
r
e
s
ume
the
b
e
ha
vior
a
l
e
nga
ge
ment
a
nd
c
ompl
e
ti
ng
the
tas
k
a
f
te
r
a
ny
int
e
r
r
upti
on
[
27
]
,
[
34]
.
c)
Af
f
e
c
ti
ve
e
nga
ge
ment
(
e
mot
ional)
Obvious
ly,
the
e
mot
ional
e
nga
ge
ment
is
de
f
ined
a
s
the
mi
r
r
or
of
a
f
f
e
c
ti
ons
a
nd
r
e
a
c
ti
on
a
mong
us
e
r
s
(
humans
)
a
nd
r
obots
w
ho
a
r
e
the
pa
r
ts
o
f
i
nter
a
c
ti
ons
whic
h
m
ight
be
a
n
int
e
r
na
l
a
nd
a
n
e
xt
e
r
na
l.
I
n
pa
r
ti
c
ular
,
the
e
mot
ional
e
nga
ge
ment
c
omp
r
is
e
s
of
s
e
ve
r
a
l
a
f
f
e
c
ti
ve
s
tate
s
,
to
na
me
f
e
w
,
e
njoym
e
nt,
mood,
the
f
e
e
li
ngs
,
a
nd
a
tt
it
ude
s
of
the
us
e
r
s
who
a
r
e
joi
ning
the
int
e
r
a
c
ti
on.
Ne
ve
r
thele
s
s
,
the
e
nthus
ias
ti
c
f
e
e
li
ng
a
nd
the
e
njoym
e
nt
a
r
e
the
domi
na
nt
a
f
f
e
c
ti
ve
s
tate
s
whic
h
ha
ve
be
e
n
inves
ti
ga
ted
in
the
va
s
t
major
it
y
of
the
done
s
tudi
e
s
[
29]
,
[
35]
.
T
he
theor
y,
that
s
a
ys
“
pos
it
ive
e
mot
ions
give
a
s
ignal
of
pur
pos
e
a
nd
e
xc
it
e
ment
to
the
br
a
in,
a
c
c
e
ler
a
ti
ng
lea
r
ning
a
nd
e
nha
nc
ing
mot
ivation”,
e
ns
ur
e
s
the
ti
ght
a
s
s
oc
iation
be
twe
e
n
the
pos
it
ive
e
mot
ions
a
nd
e
nga
ge
ment
leve
l.
He
nc
e
,
the
a
f
f
e
c
ti
ve
e
nga
ge
ment
a
mount
s
of
us
e
r
’
s
e
njoym
e
n
t
in
the
int
e
r
a
c
ti
on
e
nvir
on
ment.
Ye
t,
it
doe
s
not
c
ons
ider
a
s
indi
c
a
tor
of
the
ult
im
a
te
int
e
r
a
c
ti
on
e
f
f
e
c
t,
r
e
ga
r
dles
s
a
s
pos
it
ive
a
s
it
c
ould
be
[
24]
,
[
36]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
4,
Augus
t
2025
:
280
5
-
2814
2808
d)
S
oc
ial
e
nga
ge
ment
Ge
ne
r
a
ll
y
s
pe
a
king,
s
oc
ial
e
nga
ge
ment
is
the
wa
y
of
int
e
r
a
c
ti
on
be
twe
e
n
the
human
a
nd
it
s
e
nvir
onment
(
other
hu
man
,
tec
hnology
or
tas
k)
in
a
n
a
de
qua
te
c
ontextua
ll
y
a
ppr
oa
c
h
a
nd
s
how
s
c
ompl
ica
ted
int
e
r
na
l
dyna
mi
c
s
whic
h
indi
c
a
tes
the
oc
c
upa
ti
on
of
int
e
r
a
c
ti
on
s
tate
.
it
is
a
main
met
r
ic
f
or
mea
s
ur
e
the
human’
s
c
ognit
ive
a
nd
s
oc
io
-
e
mot
ional
s
tate
c
oll
e
c
ti
ve
ly.
Als
o,
i
t
is
de
fi
ne
d
a
s
the
qua
nti
ty
a
nd
q
ua
li
ty
of
ve
r
ba
l
a
nd
non
-
ve
r
ba
l
s
oc
ial
int
e
r
a
c
ti
on
with
r
obot
[
37]
.
I
n
HR
I
ter
m
,
s
oc
ial
e
nga
ge
ment
r
e
f
e
r
s
to
the
invol
ve
me
nt
of
the
human
with
r
obots
whic
h
ha
ve
a
f
r
iendly
a
nd
s
oc
iable
int
e
r
a
c
ti
on
c
a
pa
bil
it
y.
Additi
ona
ll
y,
it
is
a
dde
d
to
the
other
e
nga
ge
ment
c
omponents
due
to
it
s
r
e
f
e
r
e
nc
e
to
the
human
dyna
mi
c
s
a
nd
c
ons
ider
the
e
nga
ge
ment
a
s
a
n
e
xpr
e
s
s
ion
of
e
xis
t
ing
int
e
r
pe
r
s
ona
l
r
e
lations
hips
dur
ing
the
int
e
r
a
c
ti
on
[
38]
.
F
ur
ther
mor
e
,
i
t
di
f
f
e
r
s
f
r
om
other
e
nga
ge
ment
c
ompone
nts
be
c
a
us
e
of
ha
ving
di
f
f
e
r
e
nt
c
ons
c
ious
c
onc
e
ntr
icity
thr
ough
the
int
e
r
a
c
ti
on.
W
he
r
e
a
s
the
other
c
omponents
dis
r
e
ga
r
d
a
s
igni
f
ica
nt
f
a
c
tor
t
o
a
s
s
e
s
s
the
e
nga
ge
ment
dur
ing
int
e
r
a
c
ti
on
whic
h
is
the
a
c
tual
int
e
r
e
s
ti
ng
a
nd
r
e
a
dines
s
of
human
to
b
e
gin
the
int
e
r
a
c
ti
on
[
39]
.
2.
1.
2
.
E
n
gage
m
e
n
t
m
e
as
u
r
e
m
e
n
t
ap
p
r
oac
h
e
s
T
o
be
gin
with,
a
va
li
d,
r
e
li
a
ble,
a
nd
s
tur
dy
e
ng
a
ge
ment
mea
s
ur
e
ment
is
a
s
igni
f
i
c
a
nt
f
a
c
tor
f
or
de
ve
lopi
ng
a
n
int
e
r
a
c
ti
ve
r
obot
f
r
om
human’
s
pe
r
s
pe
c
ti
ve
s
ince
the
na
tur
e
of
e
nga
ge
ment
is
c
ha
ll
e
ngi
ng
to
be
mea
s
ur
e
d.
T
he
r
e
a
r
e
di
f
f
e
r
e
nt
a
ppr
oa
c
he
s
to
mea
s
ur
e
e
nga
ge
ment
s
tate
dur
ing
the
HR
I
that
ha
ve
be
e
n
f
a
ir
ly
s
tudi
e
d.
T
he
r
e
a
f
ter
,
e
a
c
h
c
a
tegor
y
ha
s
divi
de
d
int
o
s
ub
-
c
a
tegor
ies
c
ons
ider
ing
the
da
ta
modali
ti
e
s
a
nd
tec
hniques
us
e
d.
He
r
e
,
thi
s
s
tudy
high
li
ghts
a
ge
ne
r
a
l
ove
r
view
a
nd
ke
y
point
s
of
e
a
c
h
c
a
tegor
y:
a)
M
a
nua
l
mea
s
ur
e
ment
I
t
is
a
t
r
a
dit
ional
a
nd
ubiqui
tous
a
ppr
oa
c
h
to
mea
s
ur
e
the
e
nga
ge
ment
s
tate
of
us
e
r
dur
ing
HR
I
.
T
he
pr
e
domi
na
nt
tec
hniques
in
thi
s
a
ppr
oa
c
h
a
r
e
obs
e
r
va
ti
ona
l
tec
hniques
a
nd
s
e
lf
-
r
e
por
t
a
nd
que
s
ti
onna
ir
e
.
T
his
a
ppr
oa
c
h
ha
s
be
e
n
wildl
y
e
mpl
oye
d
in
va
r
ious
f
ields
;
HR
I
include
d.
I
n
c
a
s
e
of
obs
e
r
va
ti
on
meth
ods
,
the
int
e
r
a
c
ti
on’
s
a
dmi
nis
tr
a
tor
r
e
li
e
s
on
the
obs
e
r
va
ti
o
n
to
mea
s
ur
e
the
leve
l
o
f
us
e
r
’
s
e
nga
ge
ment.
T
o
n
a
me
f
e
w
of
tec
hniques
that
ha
ve
e
mpl
oye
d
in
thi
s
c
a
s
e
,
e
thogr
a
ms
a
nd
obs
e
r
va
ti
ona
l
r
a
ti
ng
s
c
a
les
[
40]
.
An
e
x
a
mpl
e
of
E
thogr
a
ms
,
video
c
oding
incor
por
a
ti
ng
obs
e
r
ve
d
e
mot
ions
that
indi
c
a
tes
to
a
na
lyzing
a
nd
labe
ll
ing
t
he
video
r
e
c
or
ding
to
c
a
tegor
ize
the
e
mot
ion
s
tate
f
or
the
in
divi
dua
ls
in
the
v
ideo
[
41]
.
On
the
other
ha
nd
,
the
e
xa
mpl
e
o
f
obs
e
r
va
ti
ona
l
r
a
ti
ng
s
c
a
les
is
obs
e
r
va
ti
ona
l
mea
s
ur
e
ment
of
e
nga
ge
ment
whic
h
us
e
s
a
n
obs
e
r
va
ti
on
c
he
c
kli
s
t
to
mea
s
ur
e
leve
l
of
e
nga
ge
ment.
On
other
ha
nd,
s
e
lf
-
r
e
por
t
a
nd
que
s
ti
onna
ir
e
i
nvolved
the
int
e
r
a
c
ti
on’
s
us
e
r
s
uc
h
a
s
us
e
r
e
nga
ge
ment
s
c
a
le
[
42]
.
T
his
a
ppr
oa
c
h
e
ndur
e
s
s
ome
dr
a
wba
c
ks
s
uc
h
a
s
the
s
ubjec
ti
vit
y
of
the
a
dmi
nis
tr
a
tor
,
ti
me
-
dis
c
r
e
pa
nc
y
is
s
ue
a
s
the
e
nga
ge
ment
is
mea
s
ur
e
d
a
f
ter
the
int
e
r
a
c
ti
on,
a
nd
lac
k
of
a
da
ptabili
ty
f
o
r
r
obo
t
dur
ing
the
int
e
r
a
c
ti
on.
b)
Automatic
mea
s
ur
e
men
t
I
n
or
de
r
to
ove
r
c
ome
the
li
mi
tations
of
manua
l
e
nga
ge
ment
mea
s
ur
e
ment,
s
e
ve
r
a
l
s
tudi
e
s
be
g
a
n
with
de
ve
lopm
e
nt
of
a
n
a
utom
a
ti
c
mea
s
ur
e
ment
methods
.
I
n
f
a
c
t,
the
idea
o
f
a
utom
a
ti
c
mea
s
ur
e
ment
of
e
nga
ge
ment
in
HR
I
is
r
e
latively
r
e
c
e
nt
then
it
ha
s
e
a
r
ne
d
mor
e
a
tt
e
nti
on
late
ly
[
6
]
.
M
os
tl
y,
the
s
tudi
e
s
uti
li
z
e
video
a
nd
a
udio
modalit
ies
of
da
ta
a
s
we
ll
a
s
the
ne
ur
ologi
c
a
l
a
nd
phys
iol
ogica
l
da
ta
f
or
mea
s
ur
e
ment
s
uc
h
a
s
he
a
r
t
r
a
te,
r
e
lative
mot
ion
index
(
R
M
I
)
,
a
nd
e
le
c
tr
oe
nc
e
pha
logr
a
m
(
EEG
)
.
How
e
ve
r
,
the
diver
s
it
y
of
s
oc
ial
r
oboti
c
s
’
a
ppli
c
a
ti
ons
ha
s
dr
a
wn
mor
e
a
tt
e
nti
on
to
wa
r
d
the
vis
ua
l
da
ta
s
ince
e
a
c
h
s
o
c
ial
r
obot
ha
s
a
buil
t
-
in
c
a
mer
a
.
Ac
c
or
dingl
y
,
the
va
s
t
major
it
y
o
f
late
s
t
s
tudi
e
s
us
e
a
c
ue
-
ba
s
e
d
a
ppr
oa
c
he
s
to
r
e
c
ognize
t
he
s
oc
ial
c
ue
s
whic
h
c
ould
m
e
a
s
ur
e
the
s
oc
ial
e
nga
ge
ment
l
e
ve
l
a
s
we
ll
a
s
other
types
of
e
nga
ge
ments
[
33]
.
A
de
ve
lopm
e
nt
of
int
e
ll
igent
r
obots
,
that
s
oc
ially
int
e
r
a
c
ted
with
human
a
nd
a
utonom
ous
ly
a
da
pted
it
s
be
ha
vior
du
r
ing
the
int
e
r
a
c
ti
on
,
r
e
quir
e
s
a
n
a
bil
it
y
to
mea
s
ur
e
the
e
nga
ge
m
e
nt
s
tate
in
p
r
ope
r
a
nd
c
onti
nuous
wa
y
whic
h
is
c
ons
ider
a
s
a
ke
y
c
h
a
ll
e
nge
f
or
s
oc
ial
r
obo
ti
c
s
r
e
s
e
a
r
c
he
r
s
.
T
he
tr
a
n
s
it
ion
to
a
utom
a
ti
c
mea
s
ur
e
ment
of
f
e
r
e
d
a
n
a
bil
it
y
to
de
ter
mi
ne
if
the
us
e
r
a
l
r
e
a
dy
e
nga
ge
d
to
the
r
obot
a
nd
wa
it
ing
f
or
it
s
r
e
s
pons
e
withi
n
the
int
e
r
a
c
ti
on
ti
me.
T
he
r
e
upon,
the
r
obo
t
c
a
n
us
e
the
e
nga
ge
ment
s
tate
to
a
da
pt
it
s
be
ha
vior
c
onve
niently
towa
r
d
e
nha
nc
ing
the
int
e
r
a
c
ti
on
outcome
.
Additi
ona
l
ly,
the
a
dva
nc
e
ment
of
mac
hine
lea
r
ning
a
nd
de
e
p
lea
r
ning
models
ha
ve
led
to
e
xpa
nd
the
im
pr
ove
ment
o
f
a
utom
a
ti
c
e
ng
a
ge
ment
mea
s
ur
e
ment
in
ter
m
o
f
a
c
c
ur
a
c
y
a
nd
c
omput
a
t
ion
a
l
ti
me
[
43
]
,
[
44]
.
F
or
the
pu
r
pos
e
of
a
utom
a
ti
c
mea
s
ur
e
men
t
r
ule
-
ba
s
e
d,
mac
hine
lea
r
ning
,
a
nd
de
e
p
lea
r
ning
a
r
e
us
e
d.
F
ir
s
tl
y
,
the
r
ule
-
ba
s
e
d
tec
hniques
that
c
ho
os
e
va
r
ious
r
ules
a
mong
the
pr
e
s
e
nc
e
of
the
ma
in
s
oc
ial
s
ignals
.
T
he
n,
e
a
c
h
r
ule
is
mea
s
ur
e
d
by
a
s
tate
ma
c
hine
that
c
a
lcula
te
the
f
inal
e
nga
ge
ment
leve
l
.
Al
s
o,
it
c
a
n
a
dopt
a
thr
e
s
hold
-
ba
s
e
d
r
ule
f
or
mea
s
ur
e
ment
pur
pos
e
.
S
e
c
ondly,
mac
hine
lea
r
ning
models
a
r
e
lar
g
e
ly
us
e
d
in
e
nga
ge
ment
mea
s
ur
e
ment
s
ince
it
e
nr
iche
s
the
de
ve
lopm
e
nt
of
HR
I
a
nd
a
f
f
e
c
ti
ve
c
omput
ing
s
tu
dies
that
a
utom
a
ti
c
a
ll
y
c
ha
r
a
c
ter
ize
the
h
uman
be
ha
vior
.
F
inally,
de
e
p
lea
r
ning
that
is
s
omew
ha
t
late
in
e
ng
a
ge
ment
mea
s
ur
e
ment
s
tudi
e
s
.
How
e
ve
r
,
both
mac
hine
lea
r
ning
a
nd
de
e
p
lea
r
ning
us
e
by
mapping
the
f
e
a
tur
e
s
of
r
a
w
da
ta
to
ge
t
the
tar
ge
t
leve
l
of
e
nga
ge
ment.
De
e
p
lea
r
ning
s
pa
r
ke
d
by
the
we
a
k
ne
s
s
of
mac
hine
lea
r
ning
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
A
n
automati
c
s
oc
ial
e
ngage
me
nt
me
as
ur
e
me
nt
dur
ing
human
-
r
obot
…
(
W
ae
l
Has
an
A
li
A
lmohamm
e
d
)
2809
models
to
de
a
l
with
high
-
dim
e
ns
ional
f
e
a
tur
e
s
a
nd
lar
ge
va
r
iations
r
a
w
da
ta
.
I
n
a
ddit
ion,
de
e
p
lea
r
ning
mi
nim
ize
s
the
c
ompl
ica
ted
mapping
in
to
a
g
r
oup
o
f
s
ub
-
mappings
[
14]
.
2.
2.
Re
lat
e
d
wor
k
I
nde
e
d
,
a
n
a
de
qua
te
wor
ks
ha
ve
be
e
n
pr
opos
e
d
f
or
e
nga
ge
ment
mea
s
ur
e
ment
in
HR
I
a
mong
dif
f
e
r
e
nt
s
c
e
na
r
ios
.
T
he
s
e
wor
k
s
ha
ve
obvious
div
e
r
s
it
y
in
c
omput
a
ti
ona
l
model
us
e
d,
da
ta
modalit
y
,
f
e
a
tur
e
s
e
ts
,
a
nd
the
number
of
e
nga
ge
ment
c
las
s
e
s
.
I
n
t
his
s
e
c
ti
on,
we
s
howc
a
s
e
of
s
e
lec
ted
wor
k
that
e
mpl
oye
d
d
if
f
e
r
e
nt
mac
hine
lea
r
ning
a
nd
de
e
p
lea
r
ning
mode
ls
f
or
a
utom
a
ti
c
e
nga
ge
ment
mea
s
ur
e
ment
in
HR
I
.
I
ni
ti
a
l
ly
,
a
dy
na
mi
c
B
a
ye
s
ian
ne
two
r
k
m
ode
l
ha
s
ut
il
ize
d
to
me
a
s
ur
e
the
e
ng
a
ge
me
nt
o
f
c
hi
ld
r
e
n
wi
th
a
uti
s
m
s
p
e
c
t
r
um
dis
o
r
de
r
(
ASD
)
i
nte
r
a
c
ti
ng
w
it
h
the
NA
O
r
obo
t.
T
he
e
va
luat
ion
da
ta
by
t
he
p
r
o
f
e
s
s
iona
l
c
a
r
e
gi
ve
r
s
us
e
d
a
s
inpu
t
t
o
t
he
m
ode
l
a
nd
the
be
s
t
pe
r
f
o
r
manc
e
o
f
t
he
m
ode
l
is
r
e
a
c
he
d
93
.
60
%
[
4
5
]
.
I
n
th
e
lik
e
manne
r
,
P
a
pa
kos
tas
e
t
a
l
.
[
15]
c
ond
uc
ted
a
m
ult
i
moda
l
mac
hine
lea
r
nin
g
a
pp
r
oa
c
h
f
or
mea
s
u
r
in
g
bina
r
y
e
nga
ge
me
nt
s
tate
f
or
c
hi
ld
r
e
n
wit
h
lea
r
ning
d
if
f
icu
l
ti
e
s
d
ur
ing
e
duc
a
ti
o
na
l
s
c
e
na
r
io
of
i
nte
r
a
c
ti
on
.
A
v
i
s
ua
l
a
n
d
a
udio
da
ta
we
r
e
c
o
ll
e
c
ted
a
nd
pr
oc
e
s
s
e
d
a
nd
th
e
Ada
B
oos
t
de
c
is
ion
t
r
e
e
e
ns
e
mbl
e
m
ode
l
ha
s
a
c
hieve
d
93.
33%
.
Add
it
i
ona
l
ly
,
E
ng
wa
ll
e
t
al
.
[
16]
p
r
op
os
e
d
a
mac
h
ine
lea
r
nin
g
m
ode
l
o
f
c
omb
ined
s
upp
o
r
t
ve
c
to
r
mac
hine
(
S
VM
)
f
or
e
n
ga
ge
me
nt
mea
s
u
r
e
me
nt
du
r
ing
HR
I
in
c
on
text
o
f
s
e
c
ond
langu
a
ge
lea
r
ni
ng
.
T
he
da
ta
c
oll
e
c
t
e
d
f
r
om
v
ideo
r
e
c
o
r
d
a
nd
t
he
highes
t
mea
s
u
r
e
ment
a
c
c
u
r
a
c
y
ha
s
a
c
h
ieve
d
is
79
.
0
0%
.
On
the
other
ha
nd,
s
ome
other
s
tudi
e
s
us
e
d
a
d
e
e
p
lea
r
ning
models
f
or
thi
s
pur
pos
e
.
F
or
ins
tanc
e
,
long
s
hor
t
-
ter
m
memor
y
(
L
S
T
M
)
-
ba
s
e
d
ne
ur
a
l
n
e
twor
k
ha
s
be
e
n
e
mp
loyed
dur
ing
un
r
e
s
tr
icte
d
c
h
il
d
-
r
obot
c
oll
a
bor
a
ti
on
f
o
r
their
e
nga
ge
ment
mea
s
ur
e
ment.
T
he
s
tudy
ha
s
be
e
n
us
e
d
the
da
ta
c
hil
d’
s
pos
e
s
a
nd
it
a
c
hieve
d
a
c
ompete
nt
a
c
c
ur
a
c
y
77.
11
%
c
ons
ider
in
g
the
dif
f
iculty
o
f
the
p
r
oblem
a
nd
in
ter
a
c
ti
on
s
c
e
n
a
r
io
[
4
]
.
Als
o,
J
a
ve
d
e
t
al
[
25
]
pr
opos
e
d
a
mul
ti
laye
r
a
nd
mul
ti
c
ha
nne
l
of
c
onvolut
ional
ne
ur
a
l
ne
twor
k
(
C
NN
)
f
or
a
utom
a
ti
c
mea
s
ur
e
ment
of
e
nga
ge
ment
in
c
hil
d
r
e
n
with
ASD
.
T
he
e
va
luation
s
howe
d
the
be
s
t
pe
r
f
or
manc
e
of
pr
opos
e
d
f
r
a
mew
or
k
is
81
.
00%
a
c
c
ur
a
c
y
us
ing
c
oll
e
c
ted
da
ta
o
f
v
ideo,
a
udio
a
nd
mot
ion
-
tr
a
c
king.
I
n
the
s
a
me
c
ontext,
a
nother
s
tudy
o
f
pr
opos
e
d
a
de
e
p
l
e
a
r
ning
models
C
NN
a
nd
L
S
T
M
.
I
t
tes
ted
the
mo
de
l
with
s
e
ve
r
a
l
vis
ua
l
da
tas
e
ts
of
dif
f
e
r
e
nt
c
ontexts
a
nd
t
he
opti
mal
pe
r
f
or
manc
e
r
e
a
c
he
d
is
89
.
00%
a
c
c
ur
a
c
y
[
14
]
.
T
a
ble
1
s
umm
a
r
ize
s
the
mentioned
-
a
bove
s
tudi
e
s
by
s
tate
d
us
e
d
model
a
nd
be
s
t
pe
r
f
or
manc
e
r
a
te.
R
e
ga
r
dles
s
s
ome
int
e
r
s
e
c
ti
on
with
other
wor
ks
,
thi
s
s
tudy
ha
s
a
n
outs
tanding
c
ontr
ibut
ion
by
e
nr
ichme
nt
the
theor
e
ti
c
a
l
l
it
e
r
a
tur
e
o
f
e
nga
ge
m
e
nt
c
onc
e
pt
a
nd
c
a
tegor
ize
s
e
nga
ge
ment
to
inde
pe
nde
nt
c
omponents
with
c
lea
r
ins
ight
a
nd
c
ha
r
a
c
ter
is
ti
c
s
.
As
we
ll
a
s
,
it
p
r
opos
e
d
a
ne
ur
a
l
ne
twor
k
c
las
s
if
ier
f
o
r
a
utom
a
ti
c
mea
s
ur
ing
mul
ti
-
c
las
s
of
s
oc
ial
e
nga
ge
ment
pa
r
ti
c
ular
ly
a
nd
i
t
wa
s
a
c
hieve
d
a
r
e
mar
ka
ble
a
c
c
ur
a
c
y
r
a
te
.
T
he
r
e
f
or
e
,
s
uc
h
r
e
s
ult
s
would
ha
ve
pr
a
c
ti
c
a
l
im
pli
c
a
ti
ons
f
or
im
p
r
ove
ment
the
int
e
r
a
c
ti
on’
s
qua
li
ty
in
di
f
f
e
r
e
nt
f
ields
.
T
a
ble
1.
S
umm
a
r
y
of
pr
e
vious
wor
k
a
nd
their
r
e
s
ul
t
R
e
f
e
r
e
nc
e
Y
e
a
r
M
ode
l
A
c
c
ur
a
c
y
(%)
[
45]
2017
D
yna
mi
c
B
a
ye
s
ia
n ne
twor
k
93.60
[
15]
2021
A
da
B
oos
t
de
c
i
s
io
n t
r
e
e
e
ns
e
mbl
e
93.33
[
16]
2022
S
V
M
79.00
[
4]
2019
L
S
T
M
-
ba
s
e
d ne
ur
a
l
ne
twor
k
77.11
[
25]
2020
C
N
N
81.00
[
14]
2020
C
N
N
a
nd L
S
T
M
89.00
3.
E
XP
E
RI
M
E
NT
AL
M
E
T
HO
D
T
he
ke
y
is
s
ue
that
ha
s
be
e
n
a
ddr
e
s
s
e
d
he
r
e
is
to
mea
s
ur
e
,
a
utom
a
ti
c
a
ll
y,
s
oc
ial
e
nga
ge
ment
s
tate
of
c
hil
dr
e
n
uti
li
z
ing
vis
ua
l
a
nd
a
udio
modalit
ies
.
T
he
e
xpe
r
im
e
nts
we
r
e
c
a
r
r
ied
out
by
us
ing
P
ytho
n
3.
10
in
Google
C
olab
e
nvir
onment.
T
he
r
e
a
r
e
s
e
ve
r
a
l
li
b
r
a
r
ies
ha
ve
be
e
n
us
e
d
dur
ing
the
e
xpe
r
im
e
nt
pr
oc
e
s
s
s
uc
h
a
s
mainly
NumP
y
a
nd
P
a
nda
s
f
or
da
ta
pr
e
pr
oc
e
s
s
ing
a
nd
ha
ndli
ng,
s
c
iki
t
-
lea
r
n
li
br
a
r
y
f
o
r
de
s
ign
a
nd
im
pleme
nt
c
las
s
if
ica
ti
on
model,
M
a
tpl
otl
ib
li
br
a
r
y
f
o
r
vis
ua
li
z
ing
the
r
e
s
ult
s
a
nd
other
s
li
br
a
r
ies
f
or
other
p
a
r
ti
c
ular
tas
ks
.
A
mul
ti
c
las
s
c
l
a
s
s
if
ica
ti
on
us
ing
M
P
L
c
la
s
s
if
ier
whe
r
e
a
s
e
a
c
h
c
las
s
r
e
pr
e
s
e
nts
a
dif
f
e
r
e
nt
s
tate
of
c
hil
d’
s
s
oc
ial
e
nga
ge
ment
a
s
de
tailed
in
ne
xt
s
e
c
ti
ons
.
F
igur
e
1
vis
ua
l
ize
the
ge
ne
r
a
l
wor
kf
low
o
f
th
e
s
tudy’
s
e
xpe
r
im
e
nt.
F
igur
e
1.
Ge
ne
r
a
l
r
e
s
e
a
r
c
h
wor
kf
low
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
4,
Augus
t
2025
:
280
5
-
2814
2810
3.
1.
Dat
as
e
t
d
e
s
c
r
ip
t
ion
F
or
s
a
ke
of
tr
a
ini
ng
a
nd
tes
ti
ng
the
pr
opos
e
d
mo
de
l,
a
publ
icly
a
va
il
a
ble
da
tas
e
t,
na
med
P
I
n
S
oR
o,
ha
s
be
e
n
us
e
d.
T
his
da
tas
e
t
wa
s
c
oll
e
c
ted
dur
ing
a
s
e
r
ies
of
unde
r
s
pe
c
if
ied
f
r
e
e
-
play
c
hil
d
-
c
hil
d
a
n
d
c
hil
d
-
r
obot
int
e
r
a
c
ti
on.
How
e
ve
r
,
it
r
e
c
or
de
d
a
n
ove
r
45
ho
ur
s
of
s
oc
ial
int
e
r
a
c
ti
ons
a
mong
45
c
hil
d
-
c
hil
d
pa
ir
s
a
nd
30
c
hil
d
-
r
obot
pa
ir
s
.
B
e
s
ides
,
it
us
e
d
a
ha
nd
-
c
ode
d
r
e
c
or
dings
oc
c
ur
r
ing
in
na
tu
r
a
l
s
oc
ial
int
e
r
a
c
ti
ons
be
twe
e
n
c
hil
dr
e
n.
I
t
ha
s
a
video
r
e
c
or
ding
,
s
ke
leta
l
inf
o
r
m
a
ti
on,
3D
r
e
c
or
d
ings
of
the
f
a
c
e
s
,
a
nd
f
ull
a
udio
r
e
c
or
ds
.
T
he
ke
y
s
tr
e
ngth
of
c
ons
ider
ing
vis
ua
l
a
nd
a
udio
da
ta
is
that
the
s
e
tup
of
int
e
r
a
c
ti
on
e
nvir
on
ment
will
be
r
e
latively
c
omf
or
table
a
nd
c
los
e
to
r
e
a
li
ty.
E
ve
nt
ua
ll
y,
the
da
tas
e
t
is
r
ich
be
c
a
us
e
of
many
c
ha
r
a
c
ter
is
ti
c
s
s
uc
h
a
s
c
ove
r
ing
a
va
s
t
r
a
nge
of
int
e
r
a
c
ti
on
s
it
ua
ti
on,
de
mons
tr
a
ti
ng
c
ompl
e
x
s
oc
ial
dyna
mi
c
s
,
na
t
ur
a
l
a
nd
or
igi
na
l
be
ha
viour
s
due
to
the
uns
pe
c
if
icity,
a
nd
we
a
lt
h
of
mul
ti
modal
int
e
r
a
c
ti
ons
[
46
]
.
P
a
r
ti
c
ul
a
r
ly,
the
da
tas
e
t
s
pe
c
if
ied
f
ive
pr
im
a
r
y
a
nd
dis
ti
nc
t
s
tate
s
of
s
oc
ial
e
nga
ge
ment
whic
h
a
r
e
:
f
ir
s
tl
y
,
s
oli
t
a
r
y
that
indi
c
a
tes
c
hil
d
dis
e
nga
ge
ment.
S
e
c
ondly,
onlooker
s
igni
f
ies
that
the
c
hil
d
is
wa
tching
the
int
e
r
a
c
ti
on
but
doe
s
not
r
e
a
ll
y
joi
n
.
T
h
ir
dly,
pa
r
a
ll
e
l
mea
ns
the
c
hil
d
joi
n
the
int
e
r
a
c
ti
on
’
s
ga
me
but
playing
s
olely.
F
our
thl
y,
a
s
s
oc
iative
r
e
f
e
r
s
that
the
c
hil
d
joi
ns
the
ga
me
wit
hout
c
oor
dination
with
other
s
a
c
ti
ve
ly
.
L
a
s
tl
y,
c
oo
p
e
r
a
ti
ve
that
s
igni
f
ies
the
c
hil
d
joi
ne
d
the
ga
me
with
a
n
or
g
a
nize
d
r
ole
a
nd
s
tar
t
s
e
ns
ing
of
tea
m
wor
k
.
3.
2.
Dat
a
p
r
e
-
p
r
oc
e
s
s
in
g
P
r
ior
to
f
e
e
d
the
ne
twor
k
by
the
da
tas
e
t,
s
e
ve
r
a
l
pr
e
-
pr
oc
e
s
s
ing
tec
hniques
ha
v
e
be
e
n
a
ppli
e
d
on
the
r
a
w
da
tas
e
t
towa
r
d
e
nha
n
c
ing
ne
twor
k’
s
pe
r
f
or
manc
e
.
I
ni
ti
a
ll
y,
di
f
f
e
r
e
nt
a
c
ti
on
ha
s
be
e
n
take
n
f
or
dif
f
e
r
e
nt
da
ta
type,
f
or
ins
tanc
e
,
c
lea
ning
the
da
ta
,
ha
ndli
ng
the
mi
s
s
ing
va
lues
,
a
nd
mana
ging
the
c
a
tegor
ica
l
f
e
a
tur
e
s
.
How
e
ve
r
,
im
ba
lanc
e
da
tas
e
t
is
a
ke
y
c
ha
ll
e
ng
e
in
e
nga
ge
ment
mea
s
ur
e
ment,
then
the
da
tas
e
t
ha
s
pa
s
s
e
d
thr
ough
s
ome
s
teps
to
be
ba
lanc
e
d
a
nd
nor
mal
ize
d.
Additi
ona
ll
y,
pr
incipa
l
c
omponent
a
na
lys
i
s
(
P
C
A)
tec
hnique
ha
s
be
e
n
uti
li
z
e
d
to
r
e
duc
e
the
high
dim
e
ns
ional
is
s
ue
in
the
r
a
w
da
tas
e
t.
On
the
other
ha
nd,
the
ir
r
e
lev
a
nt
f
e
a
tur
e
s
ha
ve
be
e
n
e
li
mi
na
ted
f
r
om
the
d
a
tas
e
t.
3.
3.
P
r
op
os
e
d
m
od
e
l
A
f
ull
y
c
onn
e
c
ted
mul
ti
laye
r
pe
r
c
e
ptr
on
(
M
LP
)
ha
s
be
e
n
us
e
d
to
our
tas
k.
T
he
f
e
a
tur
e
s
o
f
pr
e
-
pr
oc
e
s
s
e
d
da
tas
e
t
we
r
e
e
mpl
oye
d
to
tr
a
in
the
s
e
lec
ted
de
e
p
lea
r
ning
ne
ur
a
l
ne
twor
k,
M
L
P
,
to
me
a
s
ur
e
the
s
oc
ial
e
nga
ge
ment
s
tate
of
us
e
r
dur
ing
c
hil
d
-
r
obot
int
e
r
a
c
ti
on
.
M
LP
is
one
of
the
ubiqu
it
ous
f
e
e
df
or
wa
r
d
ne
ur
a
l
ne
twor
k
to
map
s
e
t
of
input
f
e
a
tur
e
s
a
nd
th
e
c
or
r
e
s
ponding
c
las
s
e
s
.
T
he
ge
ne
r
a
l
a
r
c
hit
e
c
tur
e
of
M
P
L
c
ons
is
ts
of
mul
ti
laye
r
s
with
node
s
that
f
ull
y
c
onne
c
ted
to
e
a
c
h
other
.
T
he
input
a
nd
output
laye
r
s
a
r
e
the
f
ir
s
t
a
nd
las
t
laye
r
s
s
e
que
nti
a
ll
y
in
a
ddit
ion
to
one
,
a
t
v
e
r
y
lea
s
t,
or
mul
ti
ple
h
idden
laye
r
s
in
be
twe
e
n.
M
or
e
ove
r
,
the
number
o
f
node
s
is
va
r
ying
f
or
e
ve
r
y
laye
r
in
a
c
c
or
da
nc
e
to
the
number
of
input
s
a
nd
output
s
.
T
he
M
LP
ha
s
be
e
n
s
e
lec
ted
f
or
s
oc
ial
e
nga
ge
m
e
nt
mea
s
ur
e
ment
tas
k
due
to
s
e
ve
r
a
l
r
e
a
s
ons
s
uc
h
a
s
it
s
notable
e
f
f
icie
nc
y
in
s
olvi
ng
the
non
-
li
ne
r
de
c
is
ion
bounda
r
y
a
s
we
ll
a
s
c
ompl
ica
ted
pa
tt
e
r
n
r
e
c
ognit
ion
pr
oblems
by
us
ing
non
-
li
nier
a
c
ti
va
ti
on
f
unc
ti
o
n
whic
h
e
s
s
e
nti
a
l
f
or
r
e
a
l
-
wor
ld
da
ta
s
uc
h
human
-
r
obot
e
nga
ge
ment
,
in
a
ddit
ion
to
it
s
r
obus
tnes
s
by
de
a
li
ng
with
high
-
dim
e
ns
ional
da
ta.
M
L
P
ha
s
ge
n
e
r
a
li
z
a
ti
on
a
bil
it
y
to
unt
r
a
ined
da
ta
whic
h
ove
r
c
omes
ove
r
f
it
ti
ng
a
nd
maintain
ne
w
e
xa
mpl
e
s
e
f
f
e
c
ti
ve
ly
.
I
ts
c
a
pa
bil
it
y
Als
o,
u
nli
ke
the
c
las
s
ic
mac
hine
lea
r
ning
tec
hniques
,
M
LP
ove
r
c
omes
the
f
e
a
tur
e
s
e
lec
ti
on
a
nd
f
e
a
tur
e
e
xtr
a
c
ti
on
is
s
ue
s
a
nd
de
a
ls
with
s
ubtl
e
ti
e
s
f
or
c
a
ptur
ing
the
s
oc
ial
e
nga
ge
ment
s
tate
a
c
c
ur
a
tely.
I
t
ha
s
the
a
bil
it
y
to
p
r
oc
e
s
s
the
int
r
ica
te
tape
s
tr
y
of
s
oc
ial
dy
na
mi
c
s
dur
ing
c
hil
d
-
r
obot
in
ter
a
c
ti
on
li
ke
phys
ica
l
ge
s
tur
e
s
a
nd
f
a
c
ial
e
xpr
e
s
s
ions
f
or
ins
tanc
e
.
How
e
ve
r
,
th
e
ne
twor
k
ha
s
tr
a
ined
thr
ough
the
unif
o
r
m
s
a
m
pli
ng
of
da
tas
e
t
c
a
n
mi
nim
ize
ove
r
f
it
ti
ng
a
nd
ti
me
dif
f
iculti
e
s
.
3.
4.
M
e
as
u
r
e
m
e
n
t
m
e
t
r
ics
I
n
or
de
r
to
e
va
luate
the
pe
r
f
or
manc
e
o
f
p
r
o
pos
e
d
model
f
o
r
a
utom
a
ti
c
s
oc
ial
e
nga
ge
ment
mea
s
ur
e
ment,
s
e
ve
r
a
l
a
ppr
oa
c
he
s
ha
ve
be
e
n
a
pp
li
e
d.
F
ir
s
tl
y
,
c
ompar
ing
the
c
las
s
if
ica
ti
on
r
e
s
ult
with
the
a
c
tual
c
las
s
e
s
by
c
a
lcula
ti
ng
the
a
c
c
ur
a
c
y,
pr
e
c
is
i
on
,
r
e
c
a
ll
,
a
nd
F1
-
s
c
or
e
a
s
the
f
oll
owing
e
qua
ti
on
.
Ove
r
a
ll
,
the
model
a
c
hieve
d
a
n
im
pr
e
s
s
ive
c
las
s
if
ica
ti
on
a
c
c
ur
a
c
y
r
a
te
of
94
.
85%
.
I
n
the
s
a
me
c
ontext,
p
r
e
c
is
ion
—
whic
h
de
f
ines
the
model's
pe
r
f
or
manc
e
by
c
a
lcul
a
ti
ng
the
r
a
ti
o
o
f
t
r
ue
pos
it
ives
(
T
P
)
to
the
tot
a
l
pr
e
dicte
d
pos
it
ives
(
T
P
+
f
a
ls
e
pos
it
ives
(
F
P
)
)
,
a
s
s
hown
in
(
1)
—
r
e
a
c
he
d
93.
00
%
.
L
ikew
is
e
,
r
e
c
a
ll
,
whic
h
mea
s
ur
e
s
the
r
a
ti
o
of
T
P
to
the
tot
a
l
a
c
tual
pos
it
ives
(
T
P
+
f
a
l
s
e
ne
ga
ti
ve
s
(
F
N)
)
,
a
s
de
s
c
r
ibed
in
(
2
)
,
r
e
a
c
he
d
95.
00%
.
M
e
a
nwhile,
the
F
1
-
s
c
or
e
,
de
f
ined
a
s
the
ha
r
moni
c
mea
n
of
p
r
e
c
is
ion
a
nd
r
e
c
a
ll
a
s
s
hown
in
(
3)
,
the
r
a
te
wa
s
94.
00%
.
T
a
ble
2
s
umm
a
r
ize
s
the
r
e
s
ult
s
obtaine
d
b
y
the
pr
opos
e
d
ne
twor
k
.
=
(
+
)
(
1)
=
(
+
)
(
2)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
A
n
automati
c
s
oc
ial
e
ngage
me
nt
me
as
ur
e
me
nt
dur
ing
human
-
r
obot
…
(
W
ae
l
Has
an
A
li
A
lmohamm
e
d
)
2811
1
−
=
2
×
×
+
(
3)
T
a
ble
2.
S
umm
a
r
y
of
pe
r
f
or
manc
e
mea
s
ur
e
ment
m
e
tr
ics
f
or
e
a
c
h
c
las
s
C
la
s
s
e
s
P
r
e
c
is
io
n
R
e
c
a
ll
F1
-
s
c
or
e
C
oope
r
a
ti
ve
1.00
1.00
1.00
A
s
s
oc
ia
ti
ve
0.91
1.00
0.96
P
a
r
a
ll
e
l
0.93
0.91
0.92
O
nl
ooke
r
0.91
0.89
0.92
S
ol
it
a
r
y
0.97
0.98
0.97
S
e
c
ondly,
the
model’
s
pe
r
f
or
manc
e
ha
s
be
e
n
e
va
luate
d
by
one
of
the
mos
t
us
e
d
a
ppr
oa
c
h
f
or
e
va
luating
pe
r
f
or
manc
e
of
mac
hine
lea
r
ning
a
nd
d
e
e
p
lea
r
ning
models
whic
h
is
c
onf
us
ion
matr
ix.
M
or
e
ove
r
,
c
onf
us
ion
matr
ix
is
a
c
ompr
e
he
ns
ive
pr
e
s
e
ntation
f
or
e
va
luating
mul
t
i
-
c
las
s
e
s
c
las
s
if
ier
s
’
pe
r
f
or
man
c
e
.
Als
o,
it
pr
ovides
a
vis
ua
li
z
e
d
de
piction
that
plainly
r
e
ve
a
l
ins
ight
int
o
the
number
of
pr
e
dicte
d
c
las
s
e
s
to
the
number
of
a
c
tual
c
las
s
e
s
.
How
e
ve
r
,
in
r
e
s
pe
c
t
to
our
model’
s
pe
r
f
or
manc
e
the
c
onf
us
ion
mat
r
ix
p
r
e
s
e
nts
a
br
e
a
kdown
in
de
tails
of
mea
s
ur
ing
the
s
oc
ial
e
nga
ge
ment
s
tate
of
c
hil
dr
e
n.
L
a
s
tl
y,
we
a
pply
the
r
e
c
e
iver
ope
r
a
ti
ng
c
ha
r
a
c
ter
is
ti
c
s
(
R
OC
)
to
vis
ua
li
z
e
the
mea
s
ur
e
ment
pe
r
f
or
manc
e
c
ons
ider
ing
the
c
or
r
e
c
t
a
nd
incor
r
e
c
t
mea
s
ur
e
ment
r
a
te.
R
OC
plot
ted
the
tr
a
de
be
twe
e
n
the
tr
ue
pos
it
ive
r
a
te
a
nd
f
a
ls
e
pos
it
ive
r
a
te.
F
igur
e
2
de
picte
d
the
R
OC
of
s
oc
ial
e
nga
ge
ment
mea
s
ur
e
ment
model.
F
igur
e
2
.
R
OC
of
s
oc
ial
e
nga
ge
ment
mea
s
ur
e
ment
model
4.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
I
n
the
pr
e
s
e
nt
pa
pe
r
,
the
p
r
opos
e
d
ne
ur
a
l
ne
two
r
k
met
the
e
xpe
c
tation
whe
r
e
a
s
it
ha
s
s
hown
a
r
e
mar
ka
ble
r
e
s
ult
whic
h
ve
r
if
ied
dur
ing
the
e
va
lu
a
ti
on
pha
s
e
s
a
s
de
tailed
in
the
p
r
e
vious
s
e
c
ti
on.
Als
o,
the
e
va
luation
of
ou
r
model’
s
e
xpe
r
im
e
ntal
de
mo
n
s
tr
a
ted
a
n
outper
f
o
r
manc
e
in
the
ove
r
a
ll
c
las
s
if
ica
ti
on
a
c
c
ur
a
c
y
c
ompar
ing
to
the
r
e
s
ult
of
other
wor
k
a
s
we
c
a
n
s
tate
by
s
e
e
ing
the
p
r
e
vious
wor
ks
in
T
a
ble
1.
A
notable
matter
in
the
r
e
s
ult
is
that
the
c
las
s
if
ica
ti
on
of
a
ll
c
las
s
e
s
is
c
onve
r
ge
nt,
s
ti
ll
,
ther
e
is
a
dif
f
e
r
e
nti
a
ti
on
in
the
c
oope
r
a
ti
ve
c
las
s
whic
h
may
be
a
tt
r
ibut
e
d
t
o
the
f
a
c
t
that
the
number
of
c
oope
r
a
ti
ve
c
las
s
s
a
mpl
e
s
in
the
da
tas
e
t
a
r
e
the
lea
s
t.
the
objec
ti
ve
s
of
thi
s
s
tudy
ha
ve
be
e
n
a
c
hieve
d
w
he
r
e
a
s
f
ir
s
tl
y,
a
we
ll
unde
r
s
tanding
a
nd
dis
c
us
s
ion
f
or
the
c
o
r
e
o
f
s
oc
ial
e
nga
ge
ment,
it
s
f
e
a
tur
e
s
,
a
nd
dif
f
e
r
e
nc
e
a
bout
other
c
omponents
o
f
e
nga
ge
ment
a
r
e
pr
e
s
e
nted
whic
h
r
e
f
lec
ted
on
the
s
e
tt
ing
of
mod
e
l
a
nd
the
a
c
c
ur
a
c
y’
s
im
pr
ove
ment
e
ve
ntually.
S
e
c
ondly,
de
ve
lopi
ng
a
high
a
c
c
ur
a
te
mea
s
ur
e
ment
m
ode
l
f
or
s
oc
ial
e
nga
ge
ment
s
tate
dur
ing
HR
I
.
Ye
t,
th
e
r
e
s
ult
s
highl
ight
a
li
mi
tation
in
mea
s
ur
e
ment
o
f
onlook
e
r
s
tate
that
ha
s
a
s
li
ght
de
c
li
ne
a
s
s
hown
in
t
he
r
e
c
a
ll
(
89.
00%
)
.
I
t
c
ould
be
c
a
us
e
d
by
the
f
e
a
tur
e
s
’
ove
r
l
a
p
of
thi
s
c
las
s
with
other
o
r
the
s
e
l
e
c
ted
hype
r
pa
r
a
mete
r
s
ha
ve
not
r
e
a
c
he
d
the
op
ti
mal
a
nd
a
f
f
e
c
ti
ng
the
model
pe
r
f
or
manc
e
.
Ove
r
a
ll
,
the
p
r
opos
e
d
mo
de
l
holds
pr
omi
s
e
s
f
or
s
oc
ial
e
nga
ge
ment
mea
s
ur
e
ment
in
HR
I
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
4,
Augus
t
2025
:
280
5
-
2814
2812
5.
CONC
L
USI
ON
AN
D
F
UT
UR
E
WORK
T
his
pa
pe
r
s
tudi
e
d
the
s
oc
ial
e
nga
g
e
ment
s
tate
m
e
a
s
ur
e
ment
tas
k
f
or
human,
c
hil
dr
e
n
e
xc
lus
ively,
int
e
r
a
c
ti
ng
with
a
s
oc
ial
r
obot
in
or
de
r
to
s
e
t
up
a
n
a
da
pti
ve
,
r
e
s
pons
ive,
a
nd
int
e
ll
igent
int
e
r
a
c
ti
o
n
in
r
e
a
l
ti
me
a
ppli
c
a
ti
on
whic
h
e
nha
nc
e
s
the
HR
I
a
t
las
t
pl
a
c
e
.
I
t
pr
e
s
e
nted
the
de
f
ini
ti
on
of
e
nga
ge
ment
in
H
R
I
f
ield
a
nd
dived
de
e
pe
r
to
e
a
c
h
c
omponent’
s
c
ha
r
a
c
ter
is
ti
c
s
.
How
e
ve
r
,
the
mea
s
ur
e
ment
pr
oc
e
s
s
h
a
s
be
e
n
c
ons
ider
e
d
a
s
mul
ti
c
las
s
c
las
s
if
ica
ti
on
is
s
ue
.
M
P
L
model
wa
s
us
e
d
to
tac
kle
thi
s
pr
oblem
a
nd
wa
s
a
c
hieve
d
a
dis
ti
nguis
h
r
e
s
ult
s
.
T
he
pr
opos
e
d
model
ut
i
li
z
e
d
a
mul
ti
modal
da
tas
e
t
whic
h
c
ons
is
ts
of
vis
ua
l
a
nd
a
u
dio
da
ta
f
or
t
r
a
ini
ng
a
nd
tes
ti
ng
pu
r
pos
e
.
T
he
ove
r
a
ll
a
c
c
ur
a
c
y
is
94.
85%
that
a
ppe
a
r
e
d
a
n
im
p
r
ove
ment
c
om
pa
r
ing
to
other
done
s
tudi
e
s
.
T
he
r
e
s
ult
is
pr
omi
s
ing
towa
r
d
buil
ding
a
mo
r
e
s
oc
iable
a
nd
a
da
ptable
r
obot
a
nd
l
e
ve
r
a
ge
the
int
e
r
a
c
ti
on
.
I
n
f
utu
r
e
,
we
will
wor
k
to
mea
s
ur
e
the
s
oc
ial
e
nga
ge
ment
s
tate
in
int
e
g
r
a
ted
wa
y
whic
h
mea
ns
mea
s
ur
ing
the
s
tate
s
in
be
twe
e
n
the
dis
ti
nc
t
s
tate
s
s
uc
h
a
s
(
onlooker
a
nd
pa
r
a
ll
e
l
)
o
r
(
a
s
s
oc
iative
a
nd
c
oope
r
a
ti
ve
)
a
t
the
s
a
me
ti
me
a
nd
tes
t
the
model
with
r
e
a
l
a
ppli
c
a
ti
on,
then
s
im
ulate
the
p
r
opos
e
d
model
to
the
vir
tual
r
oboti
c
s
e
nvir
onment
s
uc
h
a
s
R
OS
.
A
s
we
ll
a
s
,
ke
e
p
wo
r
king
on
im
pr
oving
the
a
c
c
ur
a
c
y
with
mor
e
da
ta
a
nd
model
pa
r
a
mete
r
s
.
AC
KNOWL
E
DGE
M
E
NT
S
T
he
a
uthor
s
gr
a
tef
ull
y
a
c
knowle
dge
the
us
e
of
s
e
r
vice
a
nd
f
a
c
il
it
ies
of
the
Ar
t
if
icia
l
I
n
telli
ge
nc
e
R
e
s
e
a
r
c
h
L
a
b
a
t
Unive
r
s
it
y
of
Ke
r
ba
la.
F
UN
DI
NG
I
NF
ORM
AT
I
ON
Author
s
a
c
knowle
dge
li
mi
ted
f
unding
p
r
ovided
by
De
pa
r
tm
e
nt
o
f
C
omput
e
r
S
c
ien
c
e
,
C
oll
e
ge
o
f
C
omput
e
r
S
c
ienc
e
a
nd
I
n
f
or
mation
T
e
c
hnology
,
U
niver
s
it
y
of
Ke
r
ba
la
.
AU
T
HO
R
CONT
RI
B
U
T
I
ONS
S
T
AT
E
M
E
N
T
T
his
jour
na
l
us
e
s
the
C
ontr
ibut
o
r
R
oles
T
a
xo
nomy
(
C
R
e
diT
)
to
r
e
c
ognize
indi
vidual
a
uthor
c
ontr
ibut
ions
,
r
e
duc
e
a
utho
r
s
hip
dis
putes
,
a
nd
f
a
c
il
it
a
te
c
oll
a
bor
a
ti
on.
Nam
e
of
Au
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
W
a
e
l
Ha
s
a
n
Ali
Almohammed
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
S
inan
Adna
n
M
uhis
n
✓
✓
✓
✓
✓
✓
✓
Z
a
hr
a
a
Abe
d
Aljas
im
M
uhis
n
✓
✓
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
li
z
a
ti
on
M
:
M
e
th
odol
ogy
So
:
So
f
twa
r
e
Va
:
Va
li
da
ti
on
Fo
:
Fo
r
ma
l
a
na
ly
s
is
I
:
I
nve
s
ti
ga
ti
on
R
:
R
e
s
our
c
e
s
D
:
D
a
ta
C
ur
a
ti
on
O
:
W
r
it
in
g
-
O
r
ig
in
a
l
D
r
a
f
t
E
:
W
r
it
in
g
-
R
e
vi
e
w
&
E
di
ti
ng
Vi
:
Vi
s
ua
li
z
a
ti
on
Su
:
Su
pe
r
vi
s
io
n
P
:
P
r
oj
e
c
t
a
dmi
ni
s
tr
a
ti
on
Fu
:
Fu
ndi
ng a
c
qui
s
it
io
n
CONF
L
I
CT
OF
I
NT
E
RE
S
T
S
T
AT
E
M
E
N
T
Author
s
s
tate
no
c
onf
li
c
t
of
int
e
r
e
s
t.
DA
T
A
AV
AI
L
A
B
I
L
I
T
Y
De
r
ived
da
ta
s
uppor
ti
ng
the
f
indi
ngs
o
f
thi
s
s
tud
y
a
r
e
a
va
il
a
ble
f
r
om
the
c
or
r
e
s
ponding
a
uthor
on
r
e
que
s
t.
RE
F
E
RE
NC
E
S
[
1]
A
.
A
la
bdul
ka
r
e
e
m,
N
.
A
lh
a
kba
ni
,
a
nd
A
.
A
l
-
N
a
f
ja
n,
“
A
s
y
s
te
ma
ti
c
r
e
vi
e
w
of
r
e
s
e
a
r
c
h
on
r
obot
-
a
s
s
is
te
d
th
e
r
a
py
f
or
c
hi
ld
r
e
n
w
it
h
a
ut
is
m,”
Se
ns
o
r
s
, vol
. 22, no. 3, J
a
n. 2022, doi:
10.3390/s
22030
944.
[
2]
N
.
G
a
s
te
ig
e
r
,
M
.
H
e
ll
ou,
a
nd
H
.
S
.
A
hn,
“
F
a
c
to
r
s
f
or
pe
r
s
on
a
li
z
a
ti
on
a
nd
lo
c
a
li
z
a
ti
on
to
opt
im
iz
e
huma
n
-
r
obot
in
te
r
a
c
ti
o
n:
a
li
te
r
a
tu
r
e
r
e
vi
e
w
,”
I
nt
e
r
nat
io
nal
J
our
nal
of
Soc
ia
l
R
obot
ic
s
, vol
. 15, no. 4, pp. 689
–
701, 2023, doi:
10.1007/s
12369
-
021
-
00811
-
8.
[
3]
C
.
L
.
v
.
S
t
r
a
te
n,
J
.
P
e
te
r
,
a
nd
R
.
K
ühne
,
“
C
hi
ld
-
r
obot
r
e
l
a
ti
ons
hi
p
f
or
ma
ti
on:
a
na
r
r
a
ti
ve
r
e
vi
e
w
of
e
mpi
r
ic
a
l
r
e
s
e
a
r
c
h,”
I
nt
e
r
nat
io
nal
J
our
nal
of
Soc
ia
l
R
obot
ic
s
, vol
. 12, no. 2, pp. 325
–
344, 2020, doi:
10.1007/s
12369
-
019
-
00569
-
0.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
A
n
automati
c
s
oc
ial
e
ngage
me
nt
me
as
ur
e
me
nt
dur
ing
human
-
r
obot
…
(
W
ae
l
Has
an
A
li
A
lmohamm
e
d
)
2813
[
4]
J
.
H
a
df
ie
ld
,
G
.
C
ha
lv
a
tz
a
ki
,
P
. K
out
r
a
s
, M
.
K
ha
m
a
s
s
i,
C
.
S
.
T
z
a
f
e
s
ta
s
,
a
nd
P
. M
a
r
a
gos
,
“
A
de
e
p
le
a
r
ni
ng
a
ppr
oa
c
h
f
or
mul
ti
-
vi
e
w
e
nga
ge
me
nt
e
s
ti
ma
ti
on
of
c
hi
ld
r
e
n
in
a
c
hi
ld
-
r
obot
jo
in
t
a
tt
e
nt
io
n
ta
s
k,”
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
I
nt
e
ll
ig
e
nt
R
obot
s
and
Sy
s
te
m
s
, pp. 12
51
–
1256, 2019, doi:
10.1109/I
R
O
S
40897.2019.
8968443.
[
5]
C
.
F
il
ip
pi
ni
a
nd
A
.
M
e
r
la
,
“
S
ys
te
ma
ti
c
r
e
vi
e
w
of
a
f
f
e
c
ti
ve
c
omput
in
g
te
c
hni
que
s
f
or
in
f
a
nt
r
obot
in
te
r
a
c
ti
on,”
I
n
te
r
nat
io
nal
J
our
nal
of
Soc
ia
l
R
obot
ic
s
, vol
. 15, no. 3, pp. 393
–
409, 2023, doi:
10.1007
/s
12369
-
023
-
00985
-
3.
[
6]
J
.
N
a
s
ir
,
B
.
B
r
uno,
M
.
C
he
to
ua
ni
,
a
nd
P
.
D
il
le
nbour
g,
“
W
h
a
t
if
s
oc
ia
l
r
obot
s
lo
ok
f
or
p
r
oduc
ti
ve
e
nga
ge
me
nt
?
:
a
ut
oma
te
d
a
s
s
e
s
s
me
nt
of
goa
l
-
c
e
nt
r
ic
e
nga
ge
me
nt
in
le
a
r
ni
ng
a
ppl
ic
a
ti
ons
,”
I
nt
e
r
nat
io
nal
J
our
nal
of
Soc
ia
l
R
obot
ic
s
,
vol
.
14,
no.
1,
pp. 55
–
71, 2022, doi:
10.1007/s
12369
-
021
-
00766
-
w.
[
7]
C
.
L
.
S
id
ne
r
,
C
.
L
e
e
,
C
.
D
.
K
id
d,
N
.
L
e
s
h,
a
nd
C
.
R
ic
h,
“
E
xpl
or
a
ti
ons
in
e
nga
ge
me
nt
f
or
huma
ns
a
nd
r
obot
s
,”
A
r
ti
fi
c
ia
l
I
nt
e
ll
ig
e
nc
e
, vol
. 166, no. 1
–
2, pp. 140
–
164, 2005, doi:
10.1016/j
.a
r
t
in
t.
2005.03.005.
[
8]
M
.
P
a
li
ga
,
“
H
uma
n
–
c
obot
in
te
r
a
c
ti
on
f
lu
e
nc
y
a
nd
c
obot
ope
r
a
to
r
s
’
jo
b
pe
r
f
or
ma
nc
e
.
T
he
me
di
a
ti
ng
r
ol
e
of
w
or
k
e
nga
g
e
me
n
t:
a
s
ur
ve
y,”
R
obot
ic
s
and A
ut
onomous
Sy
s
te
m
s
, vol
. 155, 2022, doi
:
10.1016/j
.r
obot
.2022.104191.
[
9]
S
.
N
.
K
a
r
im
a
h
a
n
d
S
.
H
a
s
e
ga
w
a
,
“
A
ut
oma
ti
c
e
nga
ge
me
nt
e
s
ti
ma
ti
on
in
s
ma
r
t
e
duc
a
ti
on/
le
a
r
ni
ng
s
e
tt
in
gs
:
a
s
ys
te
ma
ti
c
r
e
vi
e
w
of
e
nga
ge
me
nt
de
f
in
it
io
ns
,
da
ta
s
e
ts
,
a
nd
me
th
od
s
,”
Sm
a
r
t
L
e
a
r
n
in
g
E
nv
ir
onm
e
nt
s
,
vol
.
9,
no.
1,
2022,
doi
:
10.1186/s
40561
-
0
22
-
00212
-
y.
[
10]
D
.
L
a
la
,
K
.
I
noue
,
P
.
M
il
hor
a
t,
a
nd
T
.
K
a
w
a
ha
r
a
,
“
D
e
te
c
ti
o
n
of
s
oc
ia
l
s
ig
na
ls
f
or
r
e
c
ogni
z
in
g
e
nga
ge
me
nt
in
huma
n
-
r
o
bot
in
te
r
a
c
ti
on,”
ar
X
iv
-
C
om
put
e
r
Sc
ie
n
c
e
,
pp. 1
-
8,
2017.
[
11]
C
.
L
yt
r
id
is
,
C
.
B
a
z
in
a
s
,
G
.
A
.
P
a
pa
kos
ta
s
,
a
nd
V
.
K
a
bur
la
s
o
s
,
“
O
n
me
a
s
ur
in
g
e
n
ga
g
e
me
nt
le
ve
l
dur
in
g
c
hi
ld
-
r
obot
in
te
r
a
c
ti
on
in
e
duc
a
ti
on,”
A
dv
anc
e
s
i
n I
nt
e
ll
ig
e
nt
Sy
s
te
m
s
and
C
om
put
in
g
, vo
l.
1023, pp. 3
–
13, 2020, doi:
10.1007/978
-
3
-
030
-
26945
-
6_1.
[
12]
M
.
A
.
A
.
D
e
w
a
n,
M
.
M
ur
s
he
d,
a
nd
F
.
L
in
,
“
E
nga
ge
me
nt
de
te
c
ti
on
in
onl
in
e
le
a
r
ni
ng:
a
r
e
vi
e
w
,”
Sm
ar
t
L
e
ar
ni
ng
E
nv
ir
onm
e
nt
s
,
vol
. 6, no. 1, 2019, doi:
10.1186/s
40561
-
018
-
0080
-
z.
[
13]
M
.
R
.
L
im
a
,
M
.
W
a
ir
a
gka
r
,
N
.
N
a
ta
r
a
ja
n,
S
.
V
a
it
he
s
w
a
r
a
n,
a
nd
R
.
V
a
id
ya
na
th
a
n,
“
R
obot
ic
te
le
me
di
c
in
e
f
or
me
nt
a
l
he
a
lt
h:
a
mul
ti
moda
l
a
ppr
oa
c
h
to
im
pr
ove
huma
n
-
r
obot
e
nga
ge
me
nt
,”
F
r
ont
ie
r
s
in
R
obot
ic
s
and
A
I
,
vol
.
8,
2
021,
doi
:
10.3389/f
r
obt
.2021.618866.
[
14]
F
.
D
.
D
uc
he
tt
o,
P
.
B
a
xt
e
r
,
a
nd
M
.
H
a
nhe
id
e
,
“
A
r
e
you
s
ti
ll
w
it
h
me
?
C
ont
in
uous
e
ng
a
ge
me
nt
a
s
s
e
s
s
m
e
nt
f
r
om
a
r
obot
’
s
poi
n
t
of
vi
e
w
,”
F
r
ont
ie
r
s
i
n R
obot
ic
s
and A
I
, vol
. 7, 2020, doi:
10.3389/f
r
obt
.2020.00116.
[
15]
G
.
A
.
P
a
pa
kos
ta
s
e
t
al
.
,
“
E
s
ti
ma
ti
ng
c
hi
ld
r
e
n
e
ng
a
ge
me
nt
in
te
r
a
c
ti
ng
w
it
h
r
obot
s
in
s
p
e
c
ia
l
e
duc
a
ti
on
u
s
in
g
ma
c
hi
ne
le
a
r
ni
n
g,”
M
at
he
m
at
ic
al
P
r
obl
e
m
s
i
n E
ngi
ne
e
r
in
g
, vol
. 2021, 2021, doi:
10.1155/2021/
9955212.
[
16]
O
.
E
ngw
a
ll
,
R
.
C
umba
l,
J
.
L
ope
s
,
M
.
L
ju
ng,
a
nd
L
.
M
å
ns
s
on,
“
I
de
nt
if
ic
a
ti
on
of
lo
w
-
e
nga
ge
d
l
e
a
r
ne
r
s
in
r
obot
-
le
d
s
e
c
ond
la
ngua
ge
c
onve
r
s
a
ti
ons
w
it
h
a
dul
ts
,”
A
C
M
T
r
ans
ac
ti
o
ns
on
H
um
an
-
R
obot
I
nt
e
r
ac
ti
on
,
vol
.
11,
no.
2,
2
022,
doi
:
10.1145/3503799.
[
17]
A
.
S
or
r
e
nt
in
o,
G
.
M
a
nc
io
ppi
,
L
.
C
ovi
e
ll
o,
F
.
C
a
va
ll
o,
a
nd
L
.
F
io
r
in
i,
“
F
e
a
s
ib
il
it
y
s
tu
dy
on
th
e
r
ol
e
of
pe
r
s
ona
li
ty
,
e
mot
io
n,
a
nd
e
nga
ge
me
nt
in
s
oc
ia
ll
y
a
s
s
is
ti
ve
r
obot
ic
s
:
a
c
ogni
ti
ve
a
s
s
e
s
s
me
nt
s
c
e
na
r
io
,”
I
nf
or
m
at
ic
s
,
vol
.
8,
no.
2,
2021,
doi
:
10.3390/i
nf
or
ma
ti
c
s
8020023.
[
18]
J
. A
ve
li
no, L
. G
a
r
c
ia
-
M
a
r
que
s
, R
. V
e
nt
ur
a
, a
nd A
. B
e
r
na
r
di
no, “
B
r
e
a
k t
he
ic
e
:
a
s
ur
ve
y on s
oc
ia
ll
y a
w
a
r
e
e
nga
ge
me
nt
f
or
huma
n
-
r
obot
f
i
r
s
t
e
nc
ount
e
r
s
,”
I
nt
e
r
nat
io
nal
J
our
nal
of
Soc
ia
l
R
obot
i
c
s
,
vol
.
13,
no.
8,
pp.
1851
–
1877,
2021,
doi
:
10.1007/s
12369
-
020
-
00720
-
2.
[
19]
M
. A
. T
r
a
ha
n, J
. K
uo, M
. C
. C
a
r
ls
on, a
nd L
.
N
.
G
it
li
n,
“
A
s
y
s
te
ma
ti
c
r
e
vi
e
w
of
s
tr
a
te
gi
e
s
t
o f
os
te
r
a
c
ti
vi
ty
e
nga
ge
me
nt
i
n pe
r
s
ons
w
it
h de
me
nt
ia
,”
H
e
al
th
E
duc
at
io
n & B
e
hav
io
r
, vol
. 41, pp. 70S
-
83S
, 2014, doi:
10.1177/109019811
4531782.
[
20]
I
.
N
e
a
l,
S
.
H
.
J
. D
u T
oi
t,
a
nd M
.
L
ova
r
in
i,
“
T
he
u
s
e
of
t
e
c
hnol
ogy
to
pr
omot
e
me
a
ni
ngf
ul
e
ng
a
ge
me
nt
f
or
a
dul
ts
w
it
h
de
me
nt
ia
in
r
e
s
id
e
nt
ia
l
a
ge
d
c
a
r
e
:
a
s
c
opi
ng
r
e
vi
e
w
,”
I
nt
e
r
nat
io
na
l
P
s
y
c
hoge
r
ia
tr
ic
s
,
vol
.
32,
no.
8,
pp.
913
–
935,
2020,
doi
:
10.1017/S
1041610219001388.
[
21]
I
.
P
oggi
,
M
in
d,
hands
,
fa
c
e
,
and
body
:
a
goal
and
be
li
e
f
v
ie
w
of
m
ul
ti
m
odal
c
om
m
uni
c
at
io
n
,
B
e
r
li
n
,
G
e
r
ma
ny:
W
e
id
le
r
B
uc
hve
r
la
g
, 20
07
.
[
22]
K
.
D
ohe
r
ty
a
nd
G
.
D
ohe
r
ty
,
“
E
nga
ge
me
nt
in
H
C
I
:
c
onc
e
pt
io
n,
th
e
or
y
a
nd
me
a
s
ur
e
me
nt
,”
A
C
M
C
om
put
in
g
Sur
v
e
y
s
,
vol
.
51,
no. 5, 2019, doi:
10.1145/3234149.
[
23]
Y
.
F
e
ng,
E
.
I
.
B
a
r
a
kova
,
S
.
Y
u,
J
.
H
u,
a
nd
G
.
W
.
M
.
R
a
ut
e
r
b
e
r
g,
“
E
f
f
e
c
ts
of
th
e
le
ve
l
of
in
te
r
a
c
ti
vi
ty
o
f
a
s
oc
ia
l
r
obot
a
nd
th
e
r
e
s
pons
e
of
th
e
a
ugm
e
nt
e
d
r
e
a
li
ty
di
s
pl
a
y
in
c
ont
e
xt
ua
l
in
te
r
a
c
ti
ons
of
pe
opl
e
w
it
h
de
me
nt
ia
,
”
Se
n
s
or
s
,
vol
.
20,
no.
13,
pp.
1
–
12,
2020, doi:
10
.3390/s
20133771.
[
24]
Y
.
F
e
ng,
G
.
P
e
r
ugi
a
,
S
.
Y
u,
E
.
I
.
B
a
r
a
kova
,
J
.
H
u,
a
nd
G
.
W
.
M
.
R
a
ut
e
r
be
r
g,
“
C
ont
e
xt
-
e
nha
nc
e
d
hum
a
n
-
r
obot
in
te
r
a
c
ti
on:
e
xpl
or
in
g
th
e
r
ol
e
of
s
ys
te
m
in
te
r
a
c
ti
vi
ty
a
nd
mul
ti
moda
l
s
ti
mul
i
on
th
e
e
nga
ge
me
nt
of
pe
opl
e
w
it
h
de
me
nt
ia
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
So
c
ia
l
R
obot
ic
s
, vol
. 14, no. 3, pp. 807
–
826, 2022, doi:
10.1007/s
12369
-
021
-
00823
-
4.
[
25]
H
.
J
a
ve
d,
W
.
H
.
L
e
e
,
a
nd C
.
H
.
P
a
r
k,
“
C
or
r
ig
e
ndum:
to
w
a
r
d a
n
a
ut
oma
te
d
me
a
s
ur
e
of
s
oc
ia
l
e
nga
ge
m
e
nt
f
or
c
hi
ld
r
e
n
w
it
h a
ut
is
m
s
pe
c
tr
um
di
s
or
de
r
-
a
pe
r
s
ona
li
z
e
d
c
omput
a
ti
o
na
l
mode
li
ng
a
ppr
oa
c
h,”
F
r
ont
ie
r
s
in
R
obot
ic
s
and
A
I
,
vol
.
7,
2020,
doi
:
10.3389/f
r
obt
.2020.00067.
[
26]
M
.
M
ih
a
i,
C
.
N
.
A
lb
e
r
t,
V
.
C
.
M
ih
a
i,
a
nd
D
.
E
.
D
um
it
r
a
s
,
“
E
mot
io
na
l
a
nd
s
oc
ia
l
e
nga
ge
me
nt
in
th
e
e
ngl
is
h
la
ngua
ge
c
la
s
s
r
oom
f
or
hi
ghe
r
e
duc
a
ti
on s
tu
de
nt
s
i
n t
he
C
O
V
I
D
-
19 onli
ne
c
ont
e
xt
,”
Sus
ta
in
abi
li
ty
, vol
. 14, no. 8, 2022, doi
:
10.3390/s
u14084527.
[
27]
Y
.
K
im
,
S
.
B
ut
a
il
,
M
.
T
s
c
hol
l,
L
.
L
iu
,
a
nd
Y
.
W
a
ng,
“
A
n
e
xp
lo
r
a
to
r
y
a
ppr
oa
c
h
to
me
a
s
ur
in
g
c
ol
la
bor
a
ti
ve
e
nga
g
e
me
nt
in
c
hi
ld
r
obot
i
nt
e
r
a
c
ti
on,”
A
C
M
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
P
r
oc
e
e
di
ng Se
r
ie
s
, pp. 209
–
217, 2020, doi:
10.1145/3375462.3
375522.
[
28]
A
. F
. P
a
iv
a
, C
. C
unha
, G
. V
os
s
, a
nd A
. D
.
M
a
to
s
, “
T
h
e
i
nt
e
r
r
e
la
ti
ons
hi
p be
twe
e
n s
oc
i
a
l
c
onne
c
te
dne
s
s
a
nd
s
oc
ia
l
e
nga
ge
m
e
nt
a
nd
it
s
r
e
la
ti
on
w
it
h
c
ogni
ti
on:
a
s
tu
dy
us
in
g
S
H
A
R
E
da
ta
,”
A
ge
in
g
and
Soc
ie
ty
,
vol
.
43,
no.
8,
pp.
1735
–
1753,
2
023,
doi
:
10.1017/S
0144686X2100129X
.
[
29]
H
.
S
a
la
m,
O
.
C
e
li
kt
ut
a
n,
I
.
H
upont,
H
.
G
une
s
,
a
nd
M
.
C
he
to
ua
ni
,
“
F
ul
ly
a
ut
oma
ti
c
a
na
ly
s
is
of
e
nga
ge
me
nt
a
nd
it
s
r
e
la
ti
ons
hi
p
to
pe
r
s
ona
li
ty
i
n huma
n
-
r
obot
i
nt
e
r
a
c
ti
ons
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 5,
pp. 705
–
721, 2017, doi:
10.1109/AC
C
E
S
S
.2016.2614525.
[
30]
W
.
L
.
Q
.
O
ga
-
B
a
ld
w
in
a
nd
L
.
K
.
F
r
ye
r
,
“
E
nga
ge
me
nt
gr
ow
th
in
la
ngua
ge
l
e
a
r
ni
ng
c
la
s
s
r
ooms
:
a
la
te
nt
gr
ow
th
a
na
ly
s
is
of
e
nga
ge
me
nt
i
n j
a
pa
ne
s
e
e
le
me
nt
a
r
y s
c
hool
s
,
”
St
ude
nt
E
ngage
m
e
nt
i
n t
he
L
anguage
C
la
s
s
r
oom
, pp. 224
–
240, 2020.
[
31]
S
.
Y
.
B
.
H
ua
ng,
C
.
H
.
H
ua
ng, a
nd
T
.
W
.
C
h
a
ng,
“
A
ne
w
c
onc
e
pt
of
w
or
k
e
nga
ge
me
nt
th
e
or
y
in
c
ogni
ti
ve
e
ng
a
ge
me
nt
,
e
mot
io
na
l
e
nga
ge
me
nt
, a
nd phys
i
c
a
l
e
nga
ge
m
e
nt
,”
F
r
ont
ie
r
s
i
n
P
s
y
c
hol
o
gy
, vol
. 12, 2022, doi:
10.3389/f
ps
yg.2021.663440.
[
32]
A
.
H
e
ns
c
he
l,
R
.
H
or
te
n
s
iu
s
,
a
nd
E
.
S
.
C
r
os
s
,
“
S
oc
ia
l
c
ogni
ti
on
in
th
e
a
ge
of
huma
n
-
r
obot
in
te
r
a
c
ti
on,”
T
r
e
nds
in
N
e
ur
o
s
c
ie
n
c
e
s
,
vol
. 43, no. 6, pp. 373
–
384, 2020, doi:
10.1016/j
.t
in
s
.2020.03.
013.
[
33]
A
.
R
os
s
i,
M
.
R
a
ia
no,
a
nd
S
.
R
os
s
i,
“
A
f
f
e
c
ti
ve
,
c
ogni
ti
ve
a
nd
b
e
ha
vi
our
a
l
e
nga
ge
me
nt
de
te
c
ti
on
f
or
huma
n
-
r
obot
in
te
r
a
c
ti
on
i
n
a
ba
r
te
ndi
ng
s
c
e
na
r
io
,”
2021
30t
h
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
R
obot
and
H
um
an
I
nt
e
r
ac
ti
v
e
C
om
m
uni
c
at
io
n,
R
O
-
M
A
N
2021
, pp. 208
–
213, 2021, doi:
10.1109/R
O
-
M
A
N
50785.2021.9515435.
[
34]
P
.
D
a
o,
“
E
f
f
e
c
ts
of
ta
s
k
goa
l
or
ie
nt
a
ti
on
on
le
a
r
ne
r
e
nga
ge
m
e
nt
in
ta
s
k
pe
r
f
or
ma
nc
e
,”
I
R
A
L
-
I
nt
e
r
nat
io
nal
R
e
v
ie
w
of
A
ppl
ie
d
L
in
gui
s
ti
c
s
i
n L
anguage
T
e
ac
hi
ng
, vol
. 59, no. 3, pp.
315
–
334, 2021, doi:
10.1515/i
r
a
l
-
2018
-
0188.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14,
No.
4,
Augus
t
2025
:
280
5
-
2814
2814
[
35]
G
.
C
a
s
te
ll
a
no,
I
.
L
e
it
e
,
A
.
P
e
r
e
i
r
a
,
C
.
M
a
r
ti
nho,
A
.
P
a
iv
a
,
a
nd
P
.
W
.
M
c
O
w
a
n,
“
C
ont
e
xt
-
s
e
ns
it
iv
e
a
f
f
e
c
t
r
e
c
ogni
ti
on
f
or
a
r
ob
ot
ic
ga
me
c
ompa
ni
on,”
A
C
M
T
r
ans
ac
ti
on
s
on I
nt
e
r
ac
ti
v
e
I
nt
e
ll
ig
e
nt
Sy
s
te
m
s
, vol
. 4,
no. 2, 2014, doi:
10.1145/2622615.
[
36]
T
.
K
.
F
.
C
hi
u,
“
A
ppl
yi
ng
th
e
s
e
lf
-
de
te
r
mi
na
ti
on
th
e
or
y
(
S
D
T
)
to
e
xpl
a
in
s
tu
de
nt
e
nga
ge
me
nt
in
onl
in
e
le
a
r
ni
ng
dur
in
g
th
e
C
O
V
I
D
-
19
pa
nde
mi
c
,”
J
our
nal
of
R
e
s
e
ar
c
h
on
T
e
c
hnol
ogy
in
E
duc
at
io
n
,
vol
.
54,
no.
S
1,
pp.
S
14
–
S
30,
20
22,
doi
:
10.1080/15391523.2
021.1891998.
[
37]
Y
.
K
is
hi
da
a
nd
C
.
K
e
mp,
“
M
e
a
s
ur
in
g
c
hi
ld
e
nga
ge
me
nt
i
n
in
c
lu
s
iv
e
e
a
r
ly
c
hi
ld
hood
s
e
tt
in
gs
:
im
pl
ic
a
ti
ons
f
or
pr
a
c
ti
c
e
,”
A
us
tr
al
as
ia
n J
our
nal
of
E
ar
ly
C
hi
ld
hood
, vol
. 31, no. 2, pp. 14
–
19, 2006, doi:
10.117
7/
183693910603100204.
[
38]
S
.
L
e
ma
ig
na
n,
F
.
G
a
r
c
ia
,
A
.
J
a
c
q,
a
nd
P
.
D
il
le
nbour
g,
“
F
r
om
r
e
a
l
-
ti
me
a
tt
e
nt
io
n
a
s
s
e
s
s
me
nt
to
‘
w
it
h
-
me
-
ne
s
s
’
in
huma
n
-
r
o
bot
in
te
r
a
c
ti
on,”
A
C
M
/I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
H
um
an
-
R
obot
I
nt
e
r
ac
ti
on
,
pp.
157
–
164,
2016,
doi
:
10.11
09/
H
R
I
.2016.7451747.
[
39]
T
.
F
ukuda
,
Y
.
F
uka
da
,
J
.
F
a
lo
ut
,
a
nd
T
.
M
ur
phe
y,
“
H
ow
id
e
a
l
c
la
s
s
ma
te
s
pr
im
in
g
in
c
r
e
a
s
e
s
E
F
L
c
la
s
s
r
oom
pr
os
oc
ia
l
e
nga
ge
me
nt
,”
St
ude
nt
E
ngage
m
e
nt
i
n t
he
L
anguage
C
la
s
s
r
oom
, pp. 182
–
201, 202
1, doi:
10.21832/97817889
23613
-
013
.
[
40]
A
.
K
a
poor
a
nd
R
.
W
.
P
ic
a
r
d,
“
M
ul
ti
moda
l
a
f
f
e
c
t
r
e
c
ogni
ti
on
in
le
a
r
ni
ng
e
nvi
r
onme
nt
s
,”
P
r
oc
e
e
di
ngs
of
th
e
13t
h
A
C
M
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on M
ul
ti
m
e
di
a, M
M
2005
, pp. 677
–
6
82, 2005, doi:
10.1145/1101149.1
101300.
[
41]
C
.
J
one
s
,
B
.
S
ung,
a
nd
W
.
M
oyl
e
,
“
A
s
s
e
s
s
in
g
e
nga
ge
me
nt
in
pe
opl
e
w
it
h
de
me
nt
ia
:
a
ne
w
a
ppr
oa
c
h
to
a
s
s
e
s
s
m
e
nt
us
in
g
vi
de
o
a
na
ly
s
is
,”
A
r
c
hi
v
e
s
of
P
s
y
c
hi
at
r
ic
N
u
r
s
in
g
, vol
. 29, no. 6, pp. 3
77
–
382, 2015, doi:
10.1016/j
.a
pnu.2015.06.019.
[
42]
J
.
W
hi
te
hi
ll
,
Z
.
S
e
r
pe
ll
,
Y
.
C
.
L
in
,
A
.
F
os
te
r
,
a
nd
J
.
R
.
M
ove
ll
a
n,
“
T
he
f
a
c
e
s
of
e
nga
ge
m
e
nt
:
a
ut
oma
ti
c
r
e
c
ogni
ti
on
of
s
tu
de
nt
e
nga
ge
me
nt
f
r
om
f
a
c
ia
l
e
xpr
e
s
s
io
ns
,”
I
E
E
E
T
r
ans
ac
ti
ons
on
A
ff
e
c
ti
v
e
C
om
put
in
g
,
vol
.
5,
no.
1,
pp.
86
–
98,
2
014,
doi
:
10.1109/T
A
F
F
C
.2014.2316163.
[
43]
S
.
H
.
W
.
C
hua
h
a
nd
J
.
Y
u,
“
T
he
f
ut
ur
e
of
s
e
r
vi
c
e
:
th
e
pow
e
r
of
e
mot
io
n
in
huma
n
-
r
obot
in
te
r
a
c
ti
on,”
J
our
nal
of
R
e
ta
il
in
g
and
C
ons
um
e
r
Se
r
v
i
c
e
s
, vol
. 61, 2021, doi:
10.1016/j
.j
r
e
tc
ons
e
r
.202
1.102551.
[
44]
T
.
H
e
mpe
l,
L
.
D
in
ge
s
,
a
nd
A
.
A
l
-
H
a
ma
di
,
“
S
e
nt
im
e
nt
-
ba
s
e
d
e
nga
ge
me
nt
s
tr
a
te
gi
e
s
f
or
i
nt
ui
ti
ve
huma
n
-
r
obot
in
te
r
a
c
ti
on,”
P
r
oc
e
e
di
ngs
of
th
e
I
nt
e
r
nat
io
nal
J
oi
nt
C
onf
e
r
e
nc
e
on
C
o
m
put
e
r
V
is
io
n,
I
m
agi
ng
and
C
om
put
e
r
G
r
aphi
c
s
T
he
o
r
y
and
A
ppl
ic
at
io
ns
, vol
. 4, pp. 680
–
686, 2023, doi:
10.5220/0011772
900003417.
[
45]
Y
.
F
e
ng,
Q
.
J
ia
,
M
.
C
hu,
a
nd
W
.
W
e
i,
“
E
nga
ge
me
nt
e
va
lu
a
ti
on
f
or
a
ut
is
m
in
te
r
ve
nt
io
n
by
r
obot
s
ba
s
e
d
on
dyn
a
mi
c
B
a
ye
s
ia
n
ne
twor
k a
nd e
xpe
r
t
e
li
c
it
a
ti
on,”
I
E
E
E
A
c
c
e
s
s
, vol
. 5, pp. 19494
–
19504, 2017, doi:
10.1109/AC
C
E
S
S
.2017.2754291.
[
46]
S
.
L
e
ma
ig
na
n,
C
.
E
.
R
.
E
dmunds
,
E
.
S
e
nf
t,
a
nd
T
.
B
e
lp
a
e
me
,
“
T
he
P
I
nS
oR
o
da
ta
s
e
t:
s
uppor
ti
ng
th
e
d
a
ta
dr
iv
e
n
s
tu
dy
of
c
h
il
d
-
c
hi
ld
a
nd c
hi
ld
-
r
obot
s
oc
ia
l
dyna
mi
c
s
,”
P
L
oS O
N
E
, vol
. 13, no. 10, 2018, doi
:
10.1371/j
our
na
l.
pone
.0205999.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
Wa
el
Ha
s
a
n
A
l
i
A
l
m
o
ha
m
m
ed
h
o
l
d
s
a
m
as
t
er
d
e
g
ree
i
n
In
fo
rma
t
i
o
n
T
ech
n
o
l
o
g
y
fro
m
U
n
i
v
ers
i
t
i
U
t
ara
Mal
a
y
s
i
a,
Mal
ay
s
i
a
i
n
2
0
1
5
.
H
e
al
s
o
recei
v
e
d
h
i
s
B.
Sc.
(Co
mp
u
t
er
Sci
en
ce)
fro
m
U
n
i
v
ers
i
t
y
o
f
K
erb
a
l
a,
Iraq
i
n
2
0
1
2
.
H
e
i
s
cu
rre
n
t
l
y
an
as
s
i
s
t
an
t
l
ect
u
rer
at
D
e
p
art
me
n
t
o
f
Co
mp
u
t
er
Sc
i
en
c
e,
U
n
i
v
e
rs
i
t
y
o
f
K
erb
a
l
a
,
K
arb
a
l
a,
Iraq
.
H
i
s
res
earch
i
n
cl
u
d
e
s
mach
i
n
e
l
earn
i
n
g
,
affect
i
v
e
co
m
p
u
t
i
n
g
,
h
u
man
-
r
o
b
o
t
i
n
t
erac
t
i
o
n
,
an
d
n
et
w
o
r
k
s
ecu
r
i
t
y
.
H
e
can
b
e
c
o
n
t
act
e
d
at
emai
l
:
w
ael
.
h
@
u
o
k
er
b
al
a.
ed
u
.
i
q
.
Si
na
n
A
dna
n
M
uhi
s
n
h
o
l
d
s
a
m
as
t
er
d
e
g
ree
i
n
In
fo
rmat
i
o
n
T
ech
n
o
l
o
g
y
fro
m
U
n
i
v
er
s
i
t
i
U
t
ara
Mal
a
y
s
i
a,
Mal
a
y
s
i
a
i
n
2
0
1
5
.
H
e
i
s
cu
rren
t
l
y
an
as
s
i
s
t
a
n
t
l
ect
u
rer
at
t
h
e
Facu
l
t
y
o
f
B
i
o
t
ech
n
o
l
o
g
y
,
Al
-
Q
a
s
i
m
G
reen
U
n
i
v
ers
i
t
y
,
Bab
y
l
o
n
,
Iraq
.
H
i
s
re
s
earch
i
n
cl
u
d
e
s
en
t
er
p
ri
s
e
res
o
u
rce
p
l
a
n
n
i
n
g
(
E
RP),
ma
ch
i
n
e
l
earn
i
n
g
,
an
d
e
-
l
earn
i
n
g
.
H
e
ca
n
b
e
co
n
t
ac
t
ed
a
t
emai
l
:
s
i
n
an
@
u
o
q
a
s
i
m.
e
d
u
.
i
q
.
Za
hra
a
A
bed
A
l
ja
s
i
m
M
uhi
s
n
h
o
l
d
s
a
m
as
t
er
d
eg
re
e
i
n
In
fo
rmat
i
o
n
T
ech
n
o
l
o
g
y
fro
m
U
n
i
v
ers
i
t
i
U
t
ara
Mal
a
y
s
i
a,
Mal
a
y
s
i
a
i
n
2
0
1
5
.
Sh
e
i
s
c
u
rren
t
l
y
an
A
s
s
i
s
t
a
n
t
Pr
o
fes
s
o
r
a
t
Co
mp
u
t
er
Sci
e
n
ce,
Al
-
Q
as
i
m
G
reen
U
n
i
v
er
s
i
t
y
,
Bab
y
l
o
n
,
Iraq
.
H
er
res
earch
i
n
c
l
u
d
es
mac
h
i
n
e
l
earn
i
n
g
,
s
o
f
t
w
are
e
n
g
i
n
eer
i
n
g
,
k
n
o
w
l
e
d
g
e
ma
n
ag
eme
n
t
,
an
d
e
-
l
ear
n
i
n
g
.
S
h
e
can
b
e
co
n
t
ac
t
e
d
at
emai
l
:
zah
raa.
a@
u
o
q
a
s
i
m.
ed
u
.
i
q
.
Evaluation Warning : The document was created with Spire.PDF for Python.