I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Ro
bo
t
ics a
nd
Aut
o
m
a
t
io
n
(
I
J
RA)
Vo
l.
1
4
,
No
.
2
,
J
u
n
e
20
2
5
,
p
p
.
1
62
~
1
72
I
SS
N:
2722
-
2
5
8
6
,
DOI
:
1
0
.
1
1
5
9
1
/i
jr
a
.
v
14
i
2
.
pp
1
62
-
1
72
162
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
r
a
.
ia
esco
r
e.
co
m
Ca
mera
-
ba
sed si
multa
neo
us lo
ca
liza
tion a
nd ma
ppi
ng
:
metho
ds, cam
e
ra
types
,
a
nd d
eep l
e
a
rning
t
rends
Ana
k
Ag
un
g
Ng
ura
h B
a
g
us
Dwim
a
nta
ra
,
O
s
k
a
r
Na
t
a
n,
No
v
elio
P
utr
a
I
nd
a
rt
o
,
Andi
Dha
rm
a
wa
n
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
a
n
d
El
e
c
t
r
o
n
i
c
s
,
U
n
i
v
e
r
si
t
a
s
G
a
d
j
a
h
M
a
d
a
,
Y
o
g
y
a
k
a
r
t
a
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
5
,
2
0
2
4
R
ev
is
ed
Feb
2
0
,
2
0
2
5
Acc
ep
ted
Ma
r
5
,
2
0
2
5
Th
e
d
e
v
e
lo
p
m
e
n
t
o
f
sim
u
lt
a
n
e
o
u
s
l
o
c
a
li
z
a
ti
o
n
a
n
d
m
a
p
p
in
g
(S
LAM
)
tec
h
n
o
l
o
g
y
is
c
r
u
c
ial
fo
r
a
d
v
a
n
c
in
g
a
u
to
n
o
m
o
u
s
sy
ste
m
s
in
ro
b
o
ti
c
s
a
n
d
n
a
v
ig
a
ti
o
n
.
Ho
we
v
e
r,
c
a
m
e
ra
-
b
a
se
d
S
LAM
sy
ste
m
s
fa
c
e
sig
n
ifi
c
a
n
t
c
h
a
ll
e
n
g
e
s
i
n
a
c
c
u
ra
c
y
,
ro
b
u
stn
e
ss
,
a
n
d
c
o
m
p
u
tati
o
n
a
l
e
fficie
n
c
y
,
p
a
rti
c
u
larly
u
n
d
e
r
c
o
n
d
it
i
o
n
s
o
f
e
n
v
iro
n
m
e
n
tal
v
a
riab
il
i
ty
,
d
y
n
a
m
ic
sc
e
n
e
s,
a
n
d
h
a
r
d
wa
re
li
m
it
a
ti
o
n
s
.
Th
is
p
a
p
e
r
p
ro
v
id
e
s
a
c
o
m
p
re
h
e
n
siv
e
r
e
v
iew
o
f
c
a
m
e
ra
-
b
a
s
e
d
S
LAM
m
e
th
o
d
o
lo
g
ies
,
fo
c
u
si
n
g
o
n
th
e
ir
d
iffere
n
t
a
p
p
r
o
a
c
h
e
s
fo
r
p
o
se
e
stim
a
ti
o
n
,
m
a
p
re
c
o
n
str
u
c
ti
o
n
,
a
n
d
c
a
m
e
ra
ty
p
e
.
T
h
e
a
p
p
li
c
a
ti
o
n
o
f
d
e
e
p
lea
rn
i
n
g
a
lso
wil
l
b
e
d
isc
u
ss
e
d
o
n
h
o
w
it
is
e
x
p
e
c
ted
to
imp
ro
v
e
p
e
rfo
rm
a
n
c
e
.
T
h
e
o
b
jec
ti
v
e
o
f
t
h
is
p
a
p
e
r
is
to
a
d
v
a
n
c
e
t
h
e
u
n
d
e
rst
a
n
d
in
g
o
f
c
a
m
e
ra
-
b
a
s
e
d
S
LAM
s
y
ste
m
s
a
n
d
t
o
p
ro
v
id
e
a
fo
u
n
d
a
ti
o
n
f
o
r
fu
tu
re
in
n
o
v
a
t
io
n
s
i
n
r
o
b
u
s
t,
e
ff
icie
n
t
,
a
n
d
a
d
a
p
ta
b
le
S
LA
M
s
o
l
u
t
io
n
s.
A
d
d
i
t
io
n
a
l
l
y
,
it
o
ffe
rs
p
e
rti
n
e
n
t
re
fe
re
n
c
e
s
a
n
d
in
sig
h
ts
fo
r
t
h
e
d
e
sig
n
a
n
d
imp
le
m
e
n
tatio
n
o
f
n
e
x
t
-
g
e
n
e
ra
ti
o
n
S
LAM
s
y
ste
m
s a
c
ro
ss
v
a
rio
u
s a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
C
am
er
a
Dee
p
lear
n
in
g
Ma
p
r
ec
o
n
s
tr
u
ctio
n
Simu
ltan
eo
u
s
lo
ca
lizatio
n
a
n
d
m
ap
p
in
g
Vis
u
al
Od
o
m
etr
y
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Osk
ar
Nata
n
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
lectr
o
n
ics,
Un
iv
er
s
it
as Ga
d
jah
Ma
d
a
Sek
ip
Utar
a
B
u
lak
s
u
m
u
r
,
Yo
g
y
ak
ar
ta
5
5
2
8
1
,
I
n
d
o
n
esia
E
m
ail: o
s
k
ar
n
atan
@
u
g
m
.
ac
.
i
d
1.
I
NT
RO
D
UCT
I
O
N
Simu
ltan
eo
u
s
lo
ca
lizatio
n
an
d
m
ap
p
in
g
(
SLAM
)
is
a
f
u
n
d
am
en
tal
tech
n
o
lo
g
y
wid
el
y
u
s
ed
in
r
o
b
o
tics
,
au
to
n
o
m
o
u
s
v
eh
icle
s
,
an
d
o
th
er
ap
p
licatio
n
s
wh
er
e
m
ac
h
in
es
m
u
s
t
in
ter
p
r
et
an
d
n
av
ig
ate
th
eir
s
u
r
r
o
u
n
d
in
g
s
.
T
h
e
SLAM
p
r
o
ce
s
s
in
v
o
lv
es
s
im
u
ltan
eo
u
s
ly
g
en
er
atin
g
a
m
ap
o
f
an
u
n
f
a
m
iliar
en
v
ir
o
n
m
en
t
an
d
d
eter
m
in
i
n
g
th
e
d
ev
ice'
s
p
o
s
itio
n
with
in
it.
Am
o
n
g
v
ar
io
u
s
s
en
s
o
r
o
p
tio
n
s
f
o
r
S
L
AM
,
ca
m
er
as
ar
e
p
ar
ticu
lar
ly
n
o
ta
b
le
d
u
e
to
th
eir
lo
w
co
s
t,
co
m
p
ac
t
d
esig
n
,
an
d
ab
ilit
y
to
ca
p
tu
r
e
d
etail
ed
v
is
u
al
d
ata
[
1
]
.
C
am
er
a
-
b
ased
SLAM
h
as
att
r
ac
ted
s
u
b
s
tan
tial
in
ter
est
b
ec
au
s
e
it
u
tili
ze
s
v
is
u
al
in
f
o
r
m
atio
n
to
esti
m
ate
m
o
tio
n
an
d
co
n
s
tr
u
ct
m
a
p
s
,
o
f
f
er
in
g
a
c
o
s
t
-
ef
f
ec
tiv
e
alter
n
at
iv
e
to
s
en
s
o
r
s
lik
e
L
iDAR
an
d
ex
ce
llin
g
in
task
s
th
at
d
em
an
d
h
ig
h
s
p
atial
r
eso
lu
tio
n
[
2
]
.
Desp
ite
its
p
r
o
m
is
e,
v
is
u
al
S
L
AM
f
ac
es
s
ev
er
al
ch
allen
g
es,
p
ar
ticu
lar
ly
in
r
ea
l
-
wo
r
ld
a
p
p
licatio
n
s
.
Dy
n
am
ic
en
v
ir
o
n
m
en
ts
,
wh
e
r
e
o
b
jects
ar
e
co
n
s
tan
tly
m
o
v
in
g
,
ca
n
d
is
r
u
p
t
f
ea
tu
r
e
tr
ac
k
in
g
an
d
r
ed
u
ce
m
ap
p
in
g
ac
cu
r
ac
y
[
3
]
.
T
ex
tu
r
eless
s
u
r
f
ac
es,
s
u
ch
as
p
lai
n
walls
o
r
f
lo
o
r
s
,
lack
d
is
tin
g
u
is
h
ab
le
f
ea
t
u
r
es,
m
ak
in
g
it
d
if
f
icu
lt
to
ex
t
r
ac
t
an
d
m
atch
k
e
y
p
o
in
ts
[
4
]
.
Po
o
r
lig
h
tin
g
co
n
d
itio
n
s
,
s
u
ch
a
s
d
im
en
v
ir
o
n
m
e
n
ts
o
r
o
v
er
ex
p
o
s
ed
s
ce
n
es,
ca
n
d
eg
r
ad
e
im
ag
e
q
u
ality
an
d
h
in
d
er
t
h
e
s
y
s
tem
'
s
ab
ilit
y
t
o
d
etec
t
an
d
tr
ac
k
f
ea
tu
r
es
r
eliab
ly
.
T
o
o
v
er
c
o
m
e
th
ese
o
b
s
tacle
s
,
f
ea
tu
r
e
-
b
as
ed
(
in
d
ir
ec
t)
m
eth
o
d
s
[
5
]
an
d
d
ir
ec
t
m
eth
o
d
s
[
6
]
h
av
e
p
r
o
v
id
ed
a
s
o
lid
f
o
u
n
d
atio
n
f
o
r
b
u
ild
i
n
g
ac
cu
r
ate
an
d
ef
f
icien
t
m
ap
s
.
I
n
d
ir
e
ct
m
eth
o
d
s
f
o
llo
w
a
two
-
s
tep
p
r
o
ce
s
s
.
C
am
er
a
-
b
ased
SLAM
s
y
s
tem
s
ca
n
b
e
class
if
ied
in
to
th
r
ee
m
ain
ca
teg
o
r
ies
b
ased
o
n
th
e
ty
p
e
o
f
ca
m
er
a
u
s
ed
:
m
o
n
o
c
u
lar
,
s
ter
e
o
,
an
d
R
GB
-
D
s
y
s
tem
s
[
7
]
–
[
9
]
.
T
h
e
p
ip
elin
e
o
f
ca
m
er
a
-
b
ased
SLAM
g
en
er
ally
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
C
a
mera
-
b
a
s
ed
s
imu
lta
n
eo
u
s
l
o
ca
liz
a
tio
n
a
n
d
ma
p
p
in
g
…
(
A
n
a
k
A
g
u
n
g
N
g
u
r
a
h
B
a
g
u
s
Dw
ima
n
ta
r
a
)
163
co
n
s
is
ts
o
f
th
r
ee
co
r
e
s
tag
es:
p
o
s
e
esti
m
atio
n
,
lo
o
p
clo
s
u
r
e
d
etec
tio
n
,
an
d
m
ap
p
in
g
.
Po
s
e
esti
m
atio
n
in
v
o
lv
es
d
eter
m
in
in
g
th
e
ca
m
er
a’
s
p
o
s
itio
n
an
d
o
r
ie
n
tatio
n
with
in
th
e
en
v
ir
o
n
m
en
t.
L
o
o
p
clo
s
u
r
e
d
etec
tio
n
id
e
n
tifie
s
in
s
tan
ce
s
wh
er
e
th
e
ca
m
er
a
r
ev
is
its
p
r
ev
io
u
s
ly
ex
p
lo
r
e
d
ar
ea
s
,
en
ab
lin
g
th
e
s
y
s
tem
t
o
c
o
r
r
ec
t
ac
c
u
m
u
lated
er
r
o
r
s
an
d
e
n
h
an
ce
th
e
g
lo
b
al
co
n
s
is
ten
cy
o
f
th
e
esti
m
ated
tr
ajec
to
r
y
.
Ma
p
p
i
n
g
,
th
e
f
i
n
a
l
s
tag
e,
f
o
c
u
s
es
o
n
cr
ea
tin
g
a
s
tr
u
ctu
r
ed
r
ep
r
esen
tatio
n
o
f
th
e
en
v
i
r
o
n
m
e
n
t,
s
u
ch
as
a
3
D
m
ap
o
r
o
th
er
s
p
atial
m
o
d
els.
T
h
e
p
er
f
o
r
m
an
ce
o
f
SLAM
s
y
s
tem
s
is
ty
p
ically
ev
alu
ated
u
s
in
g
wid
ely
r
ec
o
g
n
ized
p
u
b
lic
d
ata
s
ets
s
u
ch
as
KI
T
T
I
[
1
0
]
,
New
C
o
lleg
e
[
1
1
]
,
T
ec
h
n
ical
Un
iv
e
r
s
ity
o
f
Mu
n
ic
h
(
T
UM
)
[
1
2
]
,
[
1
3
]
,
E
u
R
o
c
m
icr
o
ae
r
ial
v
eh
icle
(
MA
V)
[
1
4
]
,
wh
ich
s
er
v
e
as b
en
ch
m
ar
k
s
,
o
f
f
er
i
n
g
r
ea
l
-
w
o
r
l
d
d
ata
co
llected
f
r
o
m
a
v
ar
iety
o
f
s
ce
n
ar
io
s
.
T
h
e
r
ap
id
ad
v
a
n
ce
m
en
t
o
f
co
m
p
u
ter
v
is
io
n
alg
o
r
ith
m
s
h
as
s
ig
n
if
ican
tly
im
p
r
o
v
ed
ca
m
er
a
-
b
ased
SLAM
in
r
ec
en
t
y
ea
r
s
.
T
h
is
r
ev
iew
aim
s
to
p
r
o
v
id
e
a
c
o
m
p
r
eh
en
s
iv
e
o
v
e
r
v
iew
o
f
ca
m
er
a
-
b
ased
SLAM
,
f
o
cu
s
in
g
o
n
its
k
ey
co
m
p
o
n
e
n
ts
,
s
tate
-
of
-
th
e
-
ar
t
tech
n
i
q
u
e
s
,
an
d
ap
p
licatio
n
s
.
W
e
ca
teg
o
r
ize
an
d
a
n
aly
ze
ex
is
tin
g
m
eth
o
d
s
,
d
is
cu
s
s
th
eir
s
tr
en
g
th
s
an
d
lim
itatio
n
s
,
an
d
h
ig
h
lig
h
t
r
ec
en
t
tr
e
n
d
s
,
in
clu
d
in
g
th
e
in
co
r
p
o
r
atio
n
o
f
d
ee
p
lear
n
i
n
g
.
Ad
d
itio
n
ally
,
we
ad
d
r
ess
ch
allen
g
es
an
d
o
p
e
n
p
r
o
b
lem
s
in
th
e
f
ield
,
em
p
h
asizin
g
th
e
im
p
o
r
tan
ce
o
f
r
o
b
u
s
t a
n
d
s
ca
lab
le
s
o
lu
tio
n
s
f
o
r
r
ea
l
-
wo
r
ld
a
p
p
licatio
n
s
.
2.
DIFF
E
R
E
N
T
AP
P
RO
ACH
E
S
I
n
d
ir
ec
t
m
eth
o
d
s
an
d
d
ir
ec
t
m
eth
o
d
s
r
ep
r
esen
t
two
p
r
im
a
r
y
ap
p
r
o
ac
h
es
in
ca
m
er
a
-
b
as
ed
SLAM
,
ea
ch
with
its
o
wn
s
tr
en
g
th
s
a
n
d
lim
itatio
n
s
.
I
n
d
ir
ec
t
m
et
h
o
d
s
ar
e
p
ar
ticu
la
r
ly
ef
f
ec
tiv
e
in
en
v
ir
o
n
m
en
ts
r
ich
in
tex
tu
r
e.
Dir
ec
t
m
eth
o
d
s
,
in
co
n
tr
ast,
b
y
p
ass
th
e
n
ee
d
f
o
r
ex
p
licit
f
ea
tu
r
e
ex
tr
ac
tio
n
a
n
d
in
s
tead
o
p
er
ate
o
n
r
aw
p
ix
el
in
ten
s
ities
.
T
h
er
e
ar
e
also
s
ev
er
al
m
eth
o
d
s
f
o
r
r
e
co
n
s
tr
u
ctin
g
m
ap
s
,
in
clu
d
in
g
s
p
ar
s
e,
s
em
i
-
d
en
s
e,
an
d
d
en
s
e
m
eth
o
d
s
.
W
h
ile
d
en
s
e
m
eth
o
d
s
tr
y
to
u
s
e
an
d
r
e
b
u
ild
ev
er
y
p
ix
el
in
th
e
2
D
p
i
ctu
r
e
d
o
m
ain
,
s
p
ar
s
e
m
eth
o
d
s
s
im
p
ly
u
s
e
an
d
r
ec
o
n
s
tr
u
ct
a
ch
o
s
en
s
elec
tio
n
o
f
in
d
ep
en
d
en
t
p
o
in
ts
,
u
s
u
ally
c
o
r
n
er
s
.
Dir
ec
t
an
d
in
d
ir
ec
t
ar
e
n
o
t
in
ter
c
h
an
g
ea
b
l
e
with
th
e
ter
m
s
d
en
s
e
a
n
d
s
p
ar
s
e.
All
f
o
u
r
p
air
in
g
s
ar
e
ac
t
u
ally
f
ea
s
ib
le:
B
o
th
d
ir
ec
t a
n
d
s
p
ar
s
e,
as we
ll a
s
d
i
r
ec
t a
n
d
d
e
n
s
e,
in
d
ir
ec
t a
n
d
d
e
n
s
e
,
in
d
ir
ec
t a
n
d
s
p
ar
s
e
[
1
5
]
.
2
.
1
.
Dire
ct
m
et
ho
ds
T
h
e
n
u
m
b
e
r
o
f
r
ec
o
n
s
tr
u
cted
p
o
in
ts
v
ar
ies am
o
n
g
th
e
th
r
ee
ty
p
es o
f
d
ir
ec
t m
eth
o
d
s
in
SLAM
: d
en
s
e,
s
em
i
-
d
en
s
e,
an
d
s
p
ar
s
e.
T
h
es
e
m
o
d
if
icatio
n
s
s
tr
ik
e
a
co
m
p
r
o
m
is
e
b
etwe
en
tr
ad
e
-
o
f
f
s
b
etwe
en
m
ap
d
etail,
co
m
p
u
tatio
n
al
ef
f
icien
cy
,
an
d
en
v
ir
o
n
m
en
tal
r
o
b
u
s
tn
ess
.
Den
s
e
m
eth
o
d
s
u
s
e
all
o
f
th
e
p
i
x
el
in
ten
s
ity
v
alu
es
in
th
e
im
ag
e
to
r
ec
o
n
s
tr
u
ct
th
e
s
u
r
r
o
u
n
d
i
n
g
s
an
d
esti
m
ate
ca
m
er
a
m
o
tio
n
.
A
n
o
tab
le
ex
a
m
p
le
o
f
th
is
ty
p
e
is
E
last
icFu
s
io
n
by
W
h
elan
et
a
l.
[
1
6
]
,
wh
o
u
s
ed
jo
in
t
o
p
tim
iz
atio
n
,
p
h
o
to
m
etr
ic
p
o
s
e
esti
m
atio
n
an
d
g
eo
m
etr
i
c
p
o
s
e
esti
m
atio
n
.
T
h
ey
u
tili
ze
th
e
r
an
d
o
m
ized
f
er
n
e
n
co
d
in
g
[
1
7
]
f
o
r
a
p
p
ea
r
a
n
ce
-
b
ased
p
lace
r
ec
o
g
n
itio
n
an
d
f
iv
e
co
s
t
f
u
n
ctio
n
s
to
o
p
tim
iz
e
th
e
d
ef
o
r
m
atio
n
g
r
ap
h
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
g
en
e
r
ated
m
a
p
is
th
en
m
ain
tain
ed
b
y
o
p
tim
izin
g
t
h
is
d
ef
o
r
m
ati
o
n
g
r
ap
h
,
wh
ich
is
m
ad
e
u
p
o
f
a
co
llectio
n
o
f
n
o
d
es
an
d
ed
g
es
d
is
p
er
s
ed
th
r
o
u
g
h
o
u
t t
h
e
m
o
d
el
to
b
e
d
e
f
o
r
m
ed
.
Sem
i
-
d
en
s
e
m
eth
o
d
s
d
o
n
o
t
r
eb
u
ild
th
e
en
tire
s
u
r
f
ac
e.
On
e
well
-
k
n
o
wn
ex
am
p
le
th
at
s
h
o
ws
s
em
i
-
d
en
s
e
m
ap
p
in
g
ca
p
a
b
ilit
ies
in
lar
g
e
-
s
ca
le
e
n
v
ir
o
n
m
en
ts
is
lar
g
e
-
s
ca
le
d
ir
ec
t
SLAM
(
L
S
D
-
SLAM
)
[
6
]
.
T
h
e
tech
n
iq
u
e
co
m
b
in
es
f
ilter
in
g
-
b
ased
esti
m
ate
o
f
s
em
i
-
d
en
s
e
d
ep
th
m
ap
s
with
d
ir
ec
t
im
a
g
e
alig
n
m
en
t.
New
ca
m
er
a
im
ag
es
a
r
e
c
o
n
tin
u
o
u
s
ly
tr
ac
k
ed
b
y
th
e
tr
a
ck
in
g
co
m
p
o
n
e
n
t.
Fil
ter
in
g
o
v
er
s
ev
er
al
p
er
-
p
ix
el,
s
m
all
-
b
aselin
e
s
ter
eo
co
m
p
ar
is
o
n
s
in
co
n
ju
n
cti
o
n
with
in
ter
leav
e
d
s
p
atial
r
eg
u
lar
izatio
n
,
as
in
[
1
8
]
,
r
ef
in
es
d
ep
th
.
T
h
ey
id
e
n
tify
p
r
ev
io
u
s
ly
v
is
ited
ar
ea
s
u
s
in
g
f
ast
ap
p
ea
r
an
ce
-
b
ased
m
ap
p
in
g
(
FAB
MA
P)
[
1
9
]
an
d
u
tili
ze
p
o
s
e
g
r
ap
h
o
p
tim
izatio
n
to
m
in
im
iz
e
th
e
er
r
o
r
.
Fig
u
r
e
1
s
h
o
ws th
e
3
D
m
ap
r
ec
o
n
s
tr
u
ctio
n
o
f
L
S
D
-
SLAM
.
Fig
u
r
e
1
.
L
SD
-
SLAM
3
D
r
ec
o
n
s
tr
u
ctio
n
[
6
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
14
,
No
.
2
,
J
u
n
e
20
2
5
:
1
62
-
1
72
164
Den
s
e
(
o
r
s
em
i
-
d
en
s
e)
m
eth
o
d
s
,
wh
ich
u
s
u
ally
f
av
o
r
s
m
o
o
t
h
n
ess
,
cr
ea
te
a
g
eo
m
etr
ic
p
r
io
r
b
y
tak
in
g
u
s
e
o
f
th
e
co
n
n
ec
ted
n
ess
o
f
th
e
em
p
lo
y
e
d
p
ictu
r
e
r
eg
io
n
.
Ho
wev
er
,
in
th
e
s
p
ar
s
e
f
o
r
m
u
latio
n
,
g
e
o
m
etr
y
p
ar
am
eter
s
(
k
e
y
p
o
in
t
p
o
s
itio
n
s
)
ar
e
co
n
d
itio
n
ally
in
d
ep
en
d
e
n
t
g
iv
en
th
e
ca
m
e
r
a
p
o
s
es
an
d
in
tr
in
s
ics,
an
d
th
e
co
n
ce
p
t
o
f
n
eig
h
b
o
r
h
o
o
d
is
ab
s
en
t
[
1
5
]
.
T
h
ese
tech
n
iq
u
es
a
r
e
m
o
r
e
r
elian
t
o
n
th
e
p
r
esen
ce
o
f
o
b
s
er
v
ab
le
k
ey
p
o
in
ts
in
th
e
s
u
r
r
o
u
n
d
in
g
s
an
d
m
ay
r
esu
lt
in
less
d
etailed
m
ap
s
.
Fo
r
th
e
v
is
u
al
o
d
o
m
etr
y
(
VO)
ap
p
licatio
n
k
n
o
wn
as
d
ir
ec
t
s
p
ar
s
e
o
d
o
m
etr
y
(
DSO)
,
E
n
g
el
et
a
l.
[
1
5
]
ef
f
ec
tiv
ely
i
n
teg
r
ated
t
h
e
ad
v
an
tag
es
o
f
d
ir
ec
t
m
eth
o
d
s
with
th
e
ad
a
p
tab
ilit
y
o
f
s
p
ar
s
e
ap
p
r
o
ac
h
es.
T
h
ey
ac
co
m
p
lis
h
th
e
p
h
o
to
m
et
r
ic
co
u
n
ter
p
ar
t
o
f
win
d
o
wed
s
p
ar
s
e
b
u
n
d
le
ad
ju
s
tm
en
t
b
y
jo
in
tly
o
p
tim
izin
g
f
o
r
all
in
v
o
lv
ed
p
ar
am
eter
s
(
in
v
er
s
e
d
ep
th
v
alu
es,
ca
m
er
a
in
tr
in
s
ics,
an
d
ca
m
er
a
ex
tr
in
s
ic
)
.
Ad
d
itio
n
ally
,
th
e
y
m
ain
tain
th
e
g
eo
m
etr
y
r
e
p
r
esen
tatio
n
u
s
ed
b
y
p
r
ev
io
u
s
d
ir
ec
t
tec
h
n
iq
u
es,
wh
ich
is
r
ep
r
esen
tatio
n
s
o
f
3
D
p
o
in
ts
as
in
v
er
s
e
d
e
p
th
i
n
a
r
e
f
er
en
ce
f
r
am
e
.
T
h
e
ex
am
p
le
r
esu
lt
o
f
th
e
d
ir
e
ct
m
eth
o
d
is
s
h
o
wn
in
Fig
u
r
e
2
.
Fig
u
r
e
2
(
a
)
s
h
o
ws
th
e
3
D
m
ap
r
ec
o
n
s
tr
u
ctio
n
o
f
DSO.
(
a)
(
b
)
Fig
u
r
e
2
.
E
x
am
p
le
r
esu
lts
o
f
d
ir
ec
t m
eth
o
d
s
(
a)
DSO
3
D
r
ec
o
n
s
tr
u
ctio
n
[
1
5
]
an
d
(
b
)
th
e
esti
m
ated
tr
ajec
to
r
y
u
s
in
g
L
DSO
[
2
0
]
(
b
ef
o
r
e
(
r
ed
)
an
d
af
ter
(
y
ello
w)
l
o
o
p
cl
o
s
u
r
e)
Ho
wev
er
,
VO
s
u
f
f
e
r
s
f
r
o
m
th
e
cu
m
u
lativ
e
d
r
if
t
in
u
n
o
b
s
er
v
ab
le
d
eg
r
ee
s
o
f
f
r
ee
d
o
m
in
t
h
e
ab
s
en
ce
o
f
a
lo
o
p
clo
s
in
g
.
T
h
is
r
estricts
th
e
ap
p
licatio
n
to
s
h
o
r
t
-
ter
m
m
o
tio
n
esti
m
atio
n
b
ec
au
s
e
it
r
esu
lt
s
in
an
er
r
o
n
e
o
u
s
lo
n
g
-
ter
m
ca
m
er
a
tr
ajec
to
r
y
an
d
m
ap
.
A
lo
o
p
clo
s
u
r
es
m
o
d
u
le
was
a
d
d
ed
to
t
h
e
DSO
alg
o
r
ith
m
b
y
Gao
et
a
l.
[
2
0
]
.
W
h
ile
m
ain
tai
n
in
g
DSO'
s
r
esil
ien
ce
in
f
ea
tu
r
e
-
p
o
o
r
co
n
tex
ts
,
th
ey
m
o
d
if
y
its
p
o
in
t
s
elec
tio
n
ap
p
r
o
ac
h
to
p
r
i
o
r
itize
r
ec
u
r
r
in
g
co
r
n
e
r
f
ea
tu
r
es.
T
h
en
,
u
s
in
g
tr
ad
itio
n
al
B
o
W
,
th
e
ch
o
s
en
co
r
n
er
ch
ar
ac
ter
is
tics
ar
e
em
p
lo
y
ed
f
o
r
lo
o
p
cl
o
s
u
r
e
d
etec
tio
n
[
2
1
]
.
T
h
e
d
r
if
t
er
r
o
r
is
th
en
d
ec
r
e
ased
b
y
u
s
in
g
p
o
s
e
g
r
ap
h
o
p
tim
izatio
n
.
T
h
e
ef
f
e
cts
o
f
a
lo
o
p
clo
s
u
r
e
m
o
d
u
le
ar
e
d
ep
icted
in
Fig
u
r
e
2
(
b
)
.
T
ab
le
1
s
h
o
ws
th
e
co
m
p
ar
is
o
n
o
f
th
e
d
ir
ec
t m
eth
o
d
s
.
T
ab
le
1
.
T
h
e
co
m
p
ar
is
o
n
o
f
d
i
r
ec
t m
eth
o
d
s
S
LA
M
a
l
g
o
r
i
t
h
m
M
a
p
r
e
c
o
n
st
r
u
c
t
i
o
n
P
o
se
e
st
i
ma
t
i
o
n
Lo
o
p
c
l
o
s
u
r
e
s
a
n
d
g
l
o
b
a
l
m
a
p
r
e
f
i
n
e
men
t
El
a
s
t
i
c
F
u
s
i
o
n
D
e
n
se
M
i
n
i
m
i
z
e
s
t
h
e
g
e
o
me
t
r
i
c
a
n
d
p
h
o
t
o
m
e
t
r
i
c
e
r
r
o
r
s
b
e
t
w
e
e
n
t
h
e
g
l
o
b
a
l
su
r
f
a
c
e
mo
d
e
l
a
n
d
t
h
e
c
u
r
r
e
n
t
R
G
B
-
D
f
r
a
me
U
t
i
l
i
z
e
t
h
e
r
a
n
d
o
m
i
z
e
d
f
e
r
n
e
n
c
o
d
i
n
g
[
1
7
]
f
o
r
a
p
p
e
a
r
a
n
c
e
-
b
a
s
e
d
p
l
a
c
e
r
e
c
o
g
n
i
t
i
o
n
,
u
t
i
l
i
z
e
f
i
v
e
c
o
s
t
f
u
n
c
t
i
o
n
s
t
o
o
p
t
i
m
i
z
e
t
h
e
d
e
f
o
r
m
a
t
i
o
n
g
r
a
p
h
LSD
-
S
LA
M
S
e
mi
-
d
e
n
se
C
r
e
a
t
e
d
a
n
i
n
i
t
i
a
l
d
e
p
t
h
ma
p
u
si
n
g
f
r
a
me
-
to
-
f
r
a
me
m
o
t
i
o
n
e
st
i
ma
t
e
,
r
e
d
u
c
e
s
t
h
e
p
h
o
t
o
me
t
r
i
c
e
r
r
o
r
a
n
d
a
l
i
g
n
s t
h
e
r
e
f
e
r
e
n
c
e
f
r
a
m
e
w
i
t
h
t
h
e
c
u
r
r
e
n
t
f
r
a
me
I
d
e
n
t
i
f
i
e
s
p
r
e
v
i
o
u
sl
y
v
i
si
t
e
d
a
r
e
a
s
u
s
i
n
g
F
A
B
M
A
P
[
1
9
]
,
u
t
i
l
i
z
e
p
o
se
g
r
a
p
h
o
p
t
i
mi
z
a
t
i
o
n
t
o
mi
n
i
m
i
z
e
t
h
e
e
r
r
o
r
LD
S
O
S
p
a
r
se
M
a
i
n
t
a
i
n
i
n
g
t
h
e
g
e
o
m
e
t
r
y
m
o
d
e
l
u
s
e
d
b
y
o
t
h
e
r
d
i
r
e
c
t
a
p
p
r
o
a
c
h
e
s,
j
o
i
n
t
l
y
o
p
t
i
mi
z
e
f
o
r
a
l
l
r
e
l
a
t
e
d
p
a
r
a
m
e
t
e
r
s
U
t
i
l
i
z
e
c
o
r
n
e
r
f
e
a
t
u
r
e
s f
o
r
l
o
o
p
c
l
o
s
u
r
e
d
e
t
e
c
t
i
o
n
w
i
t
h
B
o
W
,
u
t
i
l
i
z
e
p
o
se
g
r
a
p
h
o
p
t
i
m
i
z
a
t
i
o
n
t
o
mi
n
i
mi
z
e
t
h
e
e
r
r
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
C
a
mera
-
b
a
s
ed
s
imu
lta
n
eo
u
s
l
o
ca
liz
a
tio
n
a
n
d
ma
p
p
in
g
…
(
A
n
a
k
A
g
u
n
g
N
g
u
r
a
h
B
a
g
u
s
Dw
ima
n
ta
r
a
)
165
2
.
2
.
I
nd
irec
t
m
et
ho
ds
I
n
d
ir
ec
t
m
eth
o
d
s
r
ely
o
n
d
ete
ctin
g
,
d
escr
ib
in
g
,
a
n
d
m
atch
i
n
g
v
is
u
al
f
ea
tu
r
es
b
etwe
en
c
o
n
s
ec
u
tiv
e
f
r
am
es,
f
o
r
in
s
tan
ce
[
5
]
,
[
2
2
]
–
[
2
5
]
.
Featu
r
e
d
etec
tio
n
a
n
d
d
escr
ip
tio
n
a
r
e
ce
n
tr
al
to
f
ea
t
u
r
e
-
b
ased
m
eth
o
d
s
.
Dete
cto
r
s
,
s
u
ch
as
f
ea
tu
r
es
f
r
o
m
ac
ce
ler
ated
s
eg
m
en
t
test
(
FAST)
,
s
p
ee
d
ed
-
u
p
r
o
b
u
s
t
f
ea
tu
r
es
(
SUR
F)
[
2
6
]
,
an
d
o
r
ien
ted
FAST
an
d
r
o
tated
B
R
I
E
F
(
OR
B
)
[
2
7
]
,
id
e
n
tif
y
f
ea
tu
r
e
p
o
in
ts
in
th
e
im
ag
e.
Fig
u
r
e
3
illu
s
tr
ates
ex
am
p
les
o
f
ex
tr
ac
te
d
f
ea
t
u
r
es
u
s
ed
i
n
f
ea
t
u
r
e
-
b
ased
m
eth
o
d
s
.
T
h
e
e
x
am
p
le
o
f
id
en
tifie
d
OR
B
f
ea
tu
r
es
in
th
e
o
u
td
o
o
r
d
ataset
[
2
8
]
a
r
e
d
is
p
lay
ed
in
Fig
u
r
e
3
(
a)
.
T
h
ese
f
ea
tu
r
es
ar
e
th
en
d
escr
ib
ed
u
s
in
g
d
escr
ip
to
r
s
th
at
a
r
e
o
f
ten
u
tili
ze
d
to
h
elp
d
etec
t
a
lo
o
p
as
in
[
5
]
,
[
2
3
]
,
[
2
9
]
.
Mo
tio
n
esti
m
atio
n
an
d
m
ap
p
i
n
g
ar
e
ac
h
iev
ed
b
y
an
aly
zin
g
th
e
c
o
r
r
esp
o
n
d
en
c
e
b
etwe
en
d
etec
ted
f
ea
tu
r
es.
Vis
u
al
o
d
o
m
etr
y
co
m
p
u
tes
th
e
r
elativ
e
m
o
tio
n
b
etwe
en
f
r
am
es,
o
f
ten
u
s
in
g
r
o
b
u
s
t
tech
n
iq
u
es
lik
e
r
a
n
d
o
m
s
am
p
le
co
n
s
en
s
u
s
(
R
ANSAC
)
[
3
0
]
to
elim
in
ate
o
u
tlier
s
,
as seen
in
[
2
3
]
,
[
2
9
]
,
[
3
1
]
.
(
a)
(
b
)
Fig
u
r
e
3
.
E
x
am
p
le
o
f
ex
tr
ac
te
d
f
ea
tu
r
es (
a)
OR
B
f
ea
tu
r
es in
o
u
td
o
o
r
d
ataset
[
2
8
]
an
d
(
b
)
o
f
SLD
ex
tr
ac
tio
n
alg
o
r
ith
m
[
2
5
]
T
h
e
in
d
i
r
ec
t
m
eth
o
d
s
ca
n
al
s
o
m
ak
e
u
s
e
o
f
lin
e
f
ea
tu
r
es
.
I
n
o
r
d
er
t
o
e
x
tr
ac
t
m
o
r
e
d
ep
en
d
ab
le
f
ea
tu
r
es
in
a
lo
w
-
tex
t
u
r
ed
en
v
ir
o
n
m
e
n
t,
L
i
et
a
l.
[
2
5
]
m
er
g
ed
p
o
in
t
a
n
d
lin
e
f
ea
t
u
r
es.
C
o
m
p
ar
ed
t
o
p
o
in
t
f
ea
tu
r
es,
lin
e
f
ea
tu
r
es
ar
e
m
o
r
e
co
m
m
o
n
in
o
u
td
o
o
r
s
ettin
g
s
an
d
ar
e
less
im
p
ac
ted
b
y
v
ar
iatio
n
s
in
illu
m
in
atio
n
.
T
h
ey
r
ef
lect
o
r
g
an
ized
e
n
v
ir
o
n
m
en
ts
m
o
r
e
ef
f
ec
tiv
ely
th
a
n
p
o
in
t
f
ea
t
u
r
e
s
an
d
p
r
o
v
id
e
m
o
r
e
im
p
o
r
tan
t
i
n
f
o
r
m
atio
n
a
b
o
u
t
th
e
g
eo
m
et
r
ic
co
n
te
n
t
o
f
an
im
ag
e.
Po
in
t
a
n
d
lin
e
f
ea
tu
r
es
ar
e
ex
tr
ac
ted
i
n
p
ar
allel
b
y
R
PL
-
SLAM
[
2
5
]
with
th
e
OR
B
m
eth
o
d
f
o
r
p
o
i
n
t
f
ea
tu
r
es
an
d
th
e
s
tr
aig
h
t
lin
e
s
eg
m
en
t
d
etec
to
r
(
SLD)
alg
o
r
ith
m
f
o
r
lin
e
s
eg
m
en
t e
x
tr
ac
tio
n
as sh
o
wn
in
Fig
u
r
e
3
(
b
)
.
I
n
d
ir
ec
t
m
eth
o
d
s
u
s
u
ally
u
tili
ze
a
s
p
ar
s
e
ap
p
r
o
ac
h
,
wh
er
e
th
e
less
d
etailed
m
ap
is
r
ec
o
n
s
tr
u
cted
.
OR
B
-
SL
AM
[
5
]
is
an
ex
am
p
le
o
f
in
d
ir
ec
t
m
eth
o
d
s
th
at
u
tili
ze
a
s
p
ar
s
e
ap
p
r
o
ac
h
,
wh
er
e
t
h
ey
ex
tr
ac
t
f
ea
tu
r
e
p
o
in
ts
u
s
in
g
th
e
OR
B
alg
o
r
i
th
m
wh
ich
ar
e
o
r
ie
n
ted
m
u
lt
is
ca
le
FAST
co
r
n
er
s
with
a
2
5
6
-
b
it
d
escr
ip
to
r
ass
o
ciate
d
.
T
h
ese
ex
tr
ac
ted
O
R
B
d
escr
ip
to
r
s
ar
e
u
tili
ze
d
to
cr
ea
te
th
e
v
o
ca
b
u
lar
y
f
o
r
th
e
p
lace
r
ec
o
g
n
itio
n
m
o
d
u
le
b
ased
o
n
a
b
ag
o
f
wo
r
d
s
(
DB
o
W
2
)
[
2
1
]
,
to
p
er
f
o
r
m
lo
o
p
d
etec
tio
n
an
d
r
e
-
lo
ca
lizat
io
n
.
A
co
v
is
ib
ilit
y
g
r
ap
h
is
co
n
s
tr
u
cte
d
alo
n
g
th
e
p
r
o
ce
s
s
,
wh
ic
h
is
b
ased
o
n
th
e
co
v
is
ib
ilit
y
in
f
o
r
m
atio
n
b
etwe
en
k
ey
f
r
am
es.
T
h
is
g
r
ap
h
is
u
tili
ze
d
to
b
u
il
d
an
ess
en
tial
g
r
a
p
h
,
i.e
.
,
a
s
p
ar
s
er
s
u
b
g
r
ap
h
o
f
t
h
e
co
v
is
ib
ilit
y
to
r
ed
u
ce
th
e
am
o
u
n
t
o
f
u
tili
ze
d
k
ey
f
r
am
es
.
An
o
p
tim
izatio
n
is
p
er
f
o
r
m
e
d
o
v
er
th
e
ess
en
tial
g
r
ap
h
u
s
i
n
g
th
e
L
ev
e
n
b
er
g
-
Ma
r
q
u
ar
d
t
alg
o
r
ith
m
im
p
lem
e
n
ted
in
g
2
o
[
3
2
]
to
m
ain
tain
g
lo
b
al
co
n
s
is
ten
cy
an
d
lo
o
p
clo
s
in
g
.
OR
B
-
SLAM
h
as
s
u
cc
ess
f
u
lly
test
ed
o
n
r
ea
l
-
wo
r
ld
d
atasets
[
1
0
]
,
[
1
1
]
,
wh
er
e
it
ca
n
h
an
d
le
lo
o
p
clo
s
u
r
e
an
d
r
e
-
l
o
ca
lizatio
n
ef
f
ec
tiv
ely
as sh
o
wn
in
Fig
u
r
e
4
.
R
eg
ar
d
in
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
R
PL
-
SLAM
,
th
e
au
th
o
r
s
claim
th
at
th
eir
p
r
o
p
o
s
e
d
m
eth
o
d
o
u
tp
er
f
o
r
m
s
th
e
R
GB
-
D
v
er
s
io
n
o
f
OR
B
-
SLAM
2
[
3
3
]
in
t
h
e
m
ajo
r
ity
o
f
s
eq
u
en
ce
s
ac
r
o
s
s
th
e
T
UM
R
GB
-
D
[
1
3
]
an
d
I
m
p
er
ial
C
o
lleg
e
L
o
n
d
o
n
an
d
Natio
n
al
U
n
iv
er
s
ity
o
f
I
r
elan
d
Ma
y
n
o
o
th
(
I
C
L
-
N
UI
M)
[
3
4
]
d
atasets
.
Ho
wev
er
,
in
ce
r
tain
d
atasets
,
th
e
p
o
s
itio
n
in
g
ac
cu
r
ac
y
o
f
R
PL
-
SLAM
d
ec
r
ea
s
es.
T
h
is
i
s
s
u
e
ar
is
es
b
ec
au
s
e,
in
im
ag
es
with
r
ich
tex
tu
r
e
i
n
f
o
r
m
atio
n
,
f
alse
p
o
s
itiv
es
in
s
tr
aig
h
t
-
lin
e
ex
tr
ac
tio
n
ca
n
o
cc
u
r
.
T
h
ese
f
alse
d
etec
tio
n
s
in
tr
o
d
u
ce
ad
d
itio
n
a
l
s
y
s
tem
n
o
is
e,
lead
in
g
to
i
n
cr
ea
s
ed
er
r
o
r
s
d
u
r
in
g
co
m
p
u
tatio
n
an
d
a
r
e
d
u
ctio
n
in
p
o
s
itio
n
in
g
ac
c
u
r
ac
y
.
T
o
ad
d
r
ess
th
is
lim
itatio
n
,
f
u
tu
r
e
r
esear
ch
will
f
o
cu
s
o
n
ex
p
lo
r
in
g
o
p
tim
izatio
n
s
tr
ateg
ies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
14
,
No
.
2
,
J
u
n
e
20
2
5
:
1
62
-
1
72
166
Fig
u
r
e
4
.
R
esu
lt o
f
OR
B
-
SLAM
in
n
ew
co
lleg
e
d
ataset
[
5
]
2
.
3
.
Sem
i
-
direct
m
et
ho
ds
I
t
is
also
f
ea
s
ib
le
to
co
m
b
in
e
d
ir
ec
t
an
d
in
d
ir
ec
t
a
p
p
r
o
ac
h
es,
as
d
em
o
n
s
tr
ated
in
[
3
5
]
–
[
3
7
]
.
A
s
em
i
-
d
ir
ec
t
v
is
u
al
o
d
o
m
etr
y
(
SVO)
was
p
r
esen
ted
b
y
Fo
s
ter
et
a
l.
[
3
3
]
th
at
co
m
b
in
es
th
e
ac
c
u
r
ac
y
a
n
d
s
p
ee
d
o
f
d
ir
ec
t
m
eth
o
d
s
with
th
e
s
u
cc
ess
cr
iter
ia
o
f
f
ea
t
u
r
e
-
b
ased
a
p
p
r
o
ac
h
es
(
k
e
y
f
r
am
e
s
elec
tio
n
,
p
a
r
allel
tr
ac
k
in
g
an
d
m
a
p
p
in
g
,
a
n
d
tr
ac
k
in
g
n
u
m
er
o
u
s
f
ea
tu
r
es).
Fo
r
m
o
tio
n
esti
m
atio
n
,
th
eir
s
em
i
-
d
ir
ec
t
m
eth
o
d
d
o
es
awa
y
with
th
e
r
eq
u
ir
em
en
t
f
o
r
ex
p
e
n
s
iv
e
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
r
eliab
le
m
atch
in
g
m
eth
o
d
s
.
At
h
ig
h
f
r
a
m
e
s
p
ee
d
s
,
th
eir
s
y
s
tem
ac
h
iev
es
s
u
b
p
ix
e
l
p
r
ec
is
io
n
b
y
wo
r
k
in
g
d
ir
ec
tl
y
with
p
ix
el
in
ten
s
ities
.
3
D
p
o
in
ts
ar
e
esti
m
ated
u
s
in
g
a
p
r
o
b
a
b
ilis
tic
m
ap
p
in
g
ap
p
r
o
ac
h
th
at
e
x
p
licitly
m
o
d
el
s
o
u
tlier
o
b
s
er
v
atio
n
s
,
r
esu
ltin
g
in
f
ewe
r
o
u
tlier
s
an
d
m
o
r
e
d
ep
e
n
d
ab
le
p
o
i
n
ts
.
I
n
s
ce
n
es
with
m
in
im
al,
r
ep
e
titi
v
e,
an
d
h
ig
h
-
f
r
e
q
u
en
c
y
te
x
tu
r
e,
r
o
b
u
s
tn
ess
is
en
h
an
ce
d
b
y
p
r
ec
is
e
an
d
h
i
g
h
f
r
am
e
-
r
ate
m
o
tio
n
esti
m
ates.
A
m
o
d
if
ied
p
ar
allel
tr
ac
k
in
g
an
d
m
ap
p
in
g
(
PTAM
)
alg
o
r
ith
m
[
3
8
]
th
at
c
an
o
p
er
ate
in
v
ast
ar
ea
s
was
u
s
ed
to
co
m
p
ar
e
th
e
p
er
f
o
r
m
a
n
ce
o
f
SVO.
PTA
M
[
3
9
]
is
o
n
e
o
f
th
e
i
n
d
ir
ec
t
m
et
h
o
d
s
u
s
ed
f
o
r
m
icr
o
ae
r
ial
v
e
h
icles
(
MA
Vs).
Acc
o
r
d
in
g
to
th
e
s
tu
d
y
,
SVO
is
a
m
o
r
e
ef
f
ec
tiv
e
o
p
tio
n
f
o
r
v
is
u
al
o
d
o
m
etr
y
in
MA
V
ap
p
licati
o
n
s
s
in
ce
it d
eliv
er
s
h
ig
h
e
r
ac
cu
r
ac
y
th
a
n
PTAM
.
3.
DIFF
E
R
E
N
T
CAM
E
RA
T
Y
P
E
S
C
am
er
a
-
b
ased
SLAM
alg
o
r
ith
m
s
em
p
lo
y
v
ar
io
u
s
ty
p
es
o
f
ca
m
er
a
s
y
s
tem
s
,
ea
ch
o
f
f
er
in
g
u
n
iq
u
e
ad
v
an
tag
es
a
n
d
lim
itatio
n
s
d
e
p
en
d
in
g
o
n
th
e
ap
p
licatio
n
an
d
en
v
ir
o
n
m
e
n
t.
Mo
n
o
cu
lar
ca
m
er
as
ar
e
am
o
n
g
th
e
m
o
s
t
co
m
m
o
n
ly
u
s
ed
d
u
e
to
th
eir
s
im
p
licity
,
af
f
o
r
d
a
b
ilit
y
,
an
d
co
m
p
ac
t
f
o
r
m
f
ac
to
r
,
an
d
ca
n
b
e
u
tili
ze
d
f
o
r
d
ir
ec
t
[
6
]
,
[
1
8
]
an
d
in
d
ir
ec
t
m
eth
o
d
s
[
5
]
,
[
7
]
,
[
2
4
]
,
[
3
8
]
,
[
3
9
]
.
Fo
r
in
d
ir
ec
t
m
eth
o
d
s
,
m
o
tio
n
b
etwe
en
two
co
n
s
ec
u
tiv
e
v
iews
is
d
eter
m
in
ed
b
y
s
o
lv
i
n
g
t
h
e
e
p
ip
o
lar
g
e
o
m
etr
y
e
q
u
atio
n
.
T
h
is
r
eq
u
i
r
es
ass
u
m
p
tio
n
s
ab
o
u
t
th
e
in
tr
in
s
ic
ca
m
er
a
p
ar
am
eter
s
.
Sin
ce
m
o
n
o
cu
lar
ca
m
er
a
s
lack
d
ep
th
in
f
o
r
m
atio
n
,
p
o
s
e
esti
m
at
io
n
o
n
ly
p
r
o
v
id
es
r
elativ
e
m
o
tio
n
,
n
o
t
ab
s
o
lu
te
s
ca
le.
T
o
a
d
d
r
ess
th
is
,
ad
d
itio
n
al
co
n
s
tr
ain
ts
o
r
ass
u
m
p
tio
n
s
,
lik
e
k
n
o
wn
o
b
ject
d
im
e
n
s
io
n
s
o
r
s
ce
n
e
r
eg
u
la
r
ities
,
ar
e
in
tr
o
d
u
c
ed
.
Ster
eo
ca
m
er
as
p
r
o
v
id
e
two
im
ag
es
f
r
o
m
lef
t
an
d
r
ig
h
t
p
er
s
p
ec
tiv
es
s
ep
ar
ated
b
y
a
f
ix
e
d
b
aselin
e.
T
h
ese
im
ag
es
ca
n
b
e
u
tili
ze
d
to
ca
p
tu
r
e
d
e
p
th
in
f
o
r
m
atio
n
b
y
tr
ian
g
u
latin
g
co
r
r
esp
o
n
d
in
g
p
o
i
n
ts
in
th
e
two
im
ag
es.
T
h
is
m
ak
es
s
ter
eo
s
y
s
tem
s
in
h
er
en
tly
m
o
r
e
r
o
b
u
s
t
in
esti
m
atin
g
s
ce
n
e
g
eo
m
etr
y
a
n
d
s
ca
le
co
m
p
ar
ed
to
m
o
n
o
cu
la
r
s
y
s
tem
s
.
Ster
e
o
ca
m
er
as
ca
n
b
e
u
tili
ze
d
f
o
r
d
ir
ec
t,
in
d
ir
ec
t,
o
r
s
em
i
-
d
ir
ec
t
m
eth
o
d
s
as
d
em
o
n
s
tr
ated
in
[
8
]
,
[
3
6
]
,
[
4
0
]
,
[
4
1
]
.
E
n
g
el
et
a
l.
[
8
]
an
d
W
an
g
et
a
l.
[
4
0
]
u
tili
ze
d
s
ter
eo
ca
m
er
a
f
o
r
d
ir
ec
t
ap
p
r
o
ac
h
es.
Ster
eo
L
SD
-
SLAM
[
8
]
u
tili
ze
b
o
th
s
tatic,
f
ix
e
d
-
b
aselin
e
s
ter
eo
an
d
tem
p
o
r
a
l,
v
ar
iab
le
-
b
aselin
e
s
ter
eo
cu
es.
T
h
eir
tech
n
iq
u
e
u
s
es
p
h
o
to
m
etr
ic
an
d
g
eo
m
et
r
ic
r
esid
u
als
at
a
s
em
i
-
d
en
s
e
s
u
b
s
et
o
f
p
ix
els
to
d
ir
ec
tly
alig
n
p
ictu
r
es.
W
h
e
n
th
er
e
is
en
o
u
g
h
in
f
o
r
m
atio
n
av
ailab
le
f
o
r
eith
er
s
tatic
o
r
tem
p
o
r
al
s
ter
eo
esti
m
atio
n
,
th
ese
p
ix
els ar
e
s
elec
ted
.
T
h
e
b
e
n
ef
its
o
f
u
s
in
g
a
s
ter
eo
ca
m
er
a
ar
e
also
h
ig
h
lig
h
te
d
in
Ster
eo
DSO
[
4
0
]
,
wh
ic
h
co
m
b
in
es
s
tatic
s
ter
eo
with
m
u
lti
-
v
iew
s
ter
eo
.
R
ath
er
th
an
r
ely
in
g
o
n
r
an
d
o
m
d
e
p
th
f
o
r
in
itializatio
n
[
6
]
,
[
1
5
]
,
[
1
8
]
,
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
C
a
mera
-
b
a
s
ed
s
imu
lta
n
eo
u
s
l
o
ca
liz
a
tio
n
a
n
d
ma
p
p
in
g
…
(
A
n
a
k
A
g
u
n
g
N
g
u
r
a
h
B
a
g
u
s
Dw
ima
n
ta
r
a
)
167
s
y
s
tem
lev
er
ag
es
d
ep
th
in
f
o
r
m
atio
n
f
r
o
m
s
tatic
s
ter
eo
m
atch
in
g
,
en
a
b
lin
g
th
e
d
ir
ec
t
ca
l
cu
latio
n
o
f
ab
s
o
lu
t
e
s
ca
le
an
d
p
r
o
v
id
i
n
g
in
itial
d
e
p
th
esti
m
ates
f
o
r
m
u
lti
-
v
iew
s
ter
eo
.
Qu
alitativ
e
an
d
q
u
a
n
ti
tativ
e
ev
alu
atio
n
s
wer
e
co
n
d
u
cted
o
n
th
e
KI
T
T
I
[
1
0
]
a
n
d
C
ity
s
ca
p
es
[
4
2
]
d
atasets
,
co
m
p
ar
in
g
th
e
r
esu
lts
with
o
th
e
r
s
ter
eo
SLAM
m
eth
o
d
s
,
s
u
ch
as
O
R
B
-
SLAM
2
[
3
3
]
an
d
Ster
eo
L
SD
-
SLAM
[
8
]
.
Acc
o
r
d
in
g
t
o
th
e
ass
es
s
m
en
ts
,
Ster
eo
DSO
o
u
tp
er
f
o
r
m
s
all
o
th
er
co
m
p
a
r
ed
ap
p
r
o
ac
h
es
in
ter
m
s
o
f
ac
cu
r
ac
y
.
Sp
ec
if
ically
,
an
an
aly
s
is
o
n
th
e
KI
T
T
I
d
ataset
s
h
o
ws
th
at
Ster
eo
DSO
o
u
tp
e
r
f
o
r
m
s
Ster
eo
O
R
B
-
SLAM
2
with
lo
o
p
clo
s
in
g
an
d
g
lo
b
al
b
u
n
d
le
ad
ju
s
tm
en
t,
ev
e
n
i
n
th
e
ab
s
en
c
e
o
f
clo
s
in
g
b
ig
lo
o
p
s
.
T
h
e
Ste
r
eo
DSO
r
esu
lt
in
t
h
e
KI
T
T
I
d
ataset
is
d
is
p
lay
ed
in
Fig
u
r
e
5
.
Fig
u
r
e
5
.
R
esu
lt o
f
Ster
eo
DS
O
o
n
s
eq
u
en
ce
0
0
o
f
th
e
K
I
T
T
I
d
ataset
[
4
0
]
R
GB
-
D
ca
m
er
as,
s
u
ch
as
Mic
r
o
s
o
f
t
Kin
ec
t
o
r
I
n
tel
R
ea
lSen
s
e,
p
r
o
v
i
d
e
b
o
th
c
o
lo
r
(
R
GB
)
an
d
d
e
p
th
(
D)
in
f
o
r
m
atio
n
d
ir
ec
tly
.
I
n
s
te
ad
of
ju
s
t
em
p
l
o
y
in
g
p
h
o
t
o
m
e
tr
ic
er
r
o
r
,
it
m
a
y
also
in
co
r
p
o
r
ate
g
eo
m
etr
ic
er
r
o
r
to
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
f
o
r
d
ir
ec
t
m
eth
o
d
s
.
B
o
th
d
ir
ec
t
a
n
d
in
d
ir
ec
t
m
et
h
o
d
s
m
ay
b
e
u
s
ed
with
R
GB
-
D
ca
m
er
as,
as
s
h
o
wn
in
[
1
6
]
,
[
2
5
]
.
T
h
e
b
e
n
ef
its
o
f
R
GB
-
D
ca
m
er
as
ar
e
d
em
o
n
s
tr
ated
in
[
2
5
]
,
[
3
3
]
.
I
t
is
p
o
s
s
ib
le
to
im
m
ed
iately
r
ec
o
v
er
th
e
3
D
in
f
o
r
m
atio
n
o
f
p
o
in
t
an
d
lin
e
ch
ar
ac
ter
is
tics
f
r
o
m
th
e
R
G
B
-
D
p
ictu
r
es
th
at
ar
e
tak
en
b
y
a
d
ep
th
ca
m
er
a.
T
h
e
p
r
ec
is
io
n
o
f
ca
m
e
r
a
lo
ca
tio
n
is
th
er
ef
o
r
e
i
n
cr
ea
s
ed
as th
e
m
at
ch
in
g
p
r
o
ce
s
s
u
s
es
3D
-
3
D
co
r
r
esp
o
n
d
e
n
ce
s
in
s
tead
o
f
th
e
2
D
-
2
D
c
o
r
r
esp
o
n
d
en
c
es f
o
u
n
d
i
n
co
n
v
en
tio
n
al
R
GB
ca
m
er
as.
4.
DE
E
P
L
E
A
RNING
AP
P
L
I
C
AT
I
O
N
S
Dee
p
lear
n
in
g
tec
h
n
iq
u
es
h
av
e
b
ee
n
m
o
r
e
a
n
d
m
o
r
e
ef
f
ec
tiv
e
as
ar
tific
ial
in
tellig
e
n
ce
h
as d
ev
elo
p
e
d
,
esp
ec
ially
in
d
o
m
ain
s
lik
e
o
b
ject
id
en
tific
atio
n
wh
er
e
th
e
y
p
r
o
v
i
d
e
n
o
ticea
b
ly
g
r
ea
ter
ac
cu
r
ac
y
[
4
3
]
.
T
h
e
f
r
o
n
t
e
n
d
o
f
co
n
v
en
tio
n
al
ca
m
er
a
-
b
ased
SLAM
tech
n
iq
u
e
s
is
b
u
ilt
o
n
m
an
u
all
y
d
esig
n
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
m
atch
i
n
g
al
g
o
r
ith
m
s
.
T
h
e
s
e
tech
n
iq
u
es,
wh
ic
h
ea
ch
h
av
e
ad
v
an
ta
g
es
an
d
d
is
ad
v
a
n
tag
es,
u
s
u
ally
em
p
l
o
y
d
escr
ip
to
r
o
r
Kan
a
d
e
-
L
u
ca
s
-
T
o
m
asi
(
KL
T
)
-
b
ased
f
ea
tu
r
e
tr
ac
k
in
g
.
Alth
o
u
g
h
KL
T
tr
ac
k
i
n
g
is
o
f
te
n
q
u
ick
er
,
it
is
le
s
s
r
es
ilien
t
to
o
cc
lu
s
io
n
s
,
ag
e
co
n
tr
ast
(
s
u
ch
as
ch
allen
g
in
g
v
is
ib
ilit
y
cir
cu
m
s
tan
c
es),
an
d
s
ig
n
if
ican
t
p
er
s
p
ec
tiv
e
s
h
if
ts
(
wh
ich
m
a
y
b
e
ca
u
s
ed
b
y
f
ast
ca
m
er
a
m
o
v
em
en
ts
)
.
L
o
n
g
e
r
-
ter
m
f
e
atu
r
e
m
o
n
ito
r
in
g
is
p
o
s
s
ib
le
u
s
in
g
d
escr
ip
to
r
-
b
ased
tr
ac
k
in
g
,
b
u
t
th
e
c
o
m
p
u
tati
o
n
al
co
s
t
is
h
ig
h
er
.
T
o
s
o
lv
e
th
is
p
r
o
b
lem
,
s
o
m
e
r
esear
ch
er
s
u
s
e
d
ee
p
lear
n
in
g
i
n
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
tr
a
ck
in
g
p
o
r
tio
n
o
f
th
e
ca
m
er
a
-
b
ased
SLAM
.
Han
et
a
l.
[
3
]
in
tr
o
d
u
ce
d
a
v
is
u
al
o
d
o
m
etr
y
s
y
s
tem
th
at
lev
er
ag
es
co
n
v
o
lu
tio
n
al
n
e
u
r
a
l
n
etwo
r
k
s
(
C
NN)
f
o
r
f
ea
tu
r
e
ex
t
r
ac
tio
n
,
s
p
ec
if
ically
u
s
in
g
th
e
Su
p
er
Po
in
t
n
etwo
r
k
[
4
4
]
.
T
h
is
ap
p
r
o
ac
h
r
ep
lace
s
tr
ad
itio
n
al
h
a
n
d
-
e
n
g
in
ee
r
e
d
f
e
atu
r
e
ex
tr
ac
tio
n
m
et
h
o
d
s
with
a
C
NN
-
b
ased
m
eth
o
d
,
wh
er
e
th
e
c
o
m
p
ar
is
o
n
is
s
h
o
wn
in
Fig
u
r
e
6
.
H
o
wev
e
r
,
th
e
r
esear
c
h
er
s
s
tated
th
a
t
th
e
s
y
s
tem
f
ailed
to
p
er
f
o
r
m
in
a
d
y
n
a
m
ic
en
v
ir
o
n
m
en
t.
Ho
wev
er
,
Ham
ess
e
et
a
l.
[
4
5
]
p
r
o
v
id
e
a
h
y
b
r
id
v
is
u
al
-
in
er
tial
o
d
o
m
etr
y
(
VI
O)
s
y
s
tem
th
at
co
m
b
in
es
a
co
n
v
en
tio
n
al
v
is
u
al
-
in
er
tial
o
p
tim
izatio
n
b
ac
k
e
n
d
with
a
d
ee
p
f
ea
tu
r
e
m
atch
i
n
g
f
r
o
n
t
en
d
.
B
ased
o
n
Su
p
e
r
Po
in
t
an
d
L
ig
h
tGlu
e
[
4
6
]
n
eu
r
al
n
etwo
r
k
s
,
th
e
au
t
h
o
r
s
cr
ea
ted
a
f
ea
tu
r
e
tr
ac
k
er
th
at
ca
n
b
e
d
ir
ec
tly
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
14
,
No
.
2
,
J
u
n
e
20
2
5
:
1
62
-
1
72
168
lin
k
ed
to
th
e
VI
NS
-
Mo
n
o
[
2
9
]
esti
m
atio
n
b
ac
k
-
en
d
.
T
h
e
s
y
s
tem
o
u
tp
er
f
o
r
m
s
th
e
s
tan
d
ar
d
VI
NS
-
Mo
n
o
,
ac
co
r
d
in
g
to
ex
ten
s
iv
e
test
in
g
o
n
Vico
n
r
o
o
m
an
d
E
u
R
o
C
[
1
4
]
m
ac
h
in
e
h
all
d
atasets
.
Fig
u
r
e
6
.
C
o
m
p
a
r
is
o
n
Su
p
e
r
Po
in
t m
atch
in
g
ab
ilit
y
to
o
t
h
er
t
r
ad
itio
n
al
alg
o
r
ith
m
s
[
4
4
]
T
r
ad
itio
n
al
SLAM
s
y
s
tem
s
t
y
p
ically
r
ely
o
n
lo
w
-
lev
el
g
eo
m
etr
ic
f
ea
tu
r
es,
wh
ich
ca
n
r
esu
lt
in
f
ailu
r
es
in
r
ec
o
g
n
izin
g
lo
o
p
cl
o
s
u
r
es
in
en
v
ir
o
n
m
en
ts
with
r
ep
etitiv
e
o
r
u
n
clea
r
v
is
u
al
in
f
o
r
m
atio
n
.
Fo
r
lo
o
p
clo
s
u
r
e
d
etec
tio
n
,
s
ev
er
al
m
et
h
o
d
s
also
h
av
e
b
ee
n
p
r
o
p
o
s
ed
to
im
p
r
o
v
e
SLAM
p
er
f
o
r
m
a
n
ce
.
C
h
en
et
a
l.
[
4
7
]
p
r
o
p
o
s
ed
a
m
eth
o
d
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
o
f
tr
a
d
itio
n
al
OR
B
-
SLAM
2
[
3
3
]
b
y
in
co
r
p
o
r
atin
g
s
em
an
tic
in
f
o
r
m
atio
n
th
r
o
u
g
h
th
e
Ma
s
k
R
-
C
NN
m
o
d
el.
T
h
e
Ma
s
k
R
-
C
NN
m
o
d
el
d
etec
ts
o
b
jects
in
th
e
im
ag
e
,
p
r
o
v
id
es
s
em
an
tic
lab
els,
an
d
g
iv
es
a
h
ig
h
-
q
u
ality
s
eg
m
e
n
tatio
n
r
esu
lt
to
th
e
o
b
ject.
On
th
e
o
th
er
h
an
d
,
Dai
et
a
l.
[
4
8
]
u
tili
ze
d
R
esn
et3
4
to
d
etec
t lo
o
p
cl
o
s
u
r
es.
Fo
r
h
an
d
lin
g
th
e
p
r
o
b
lem
o
f
p
er
f
o
r
m
in
g
in
a
d
y
n
am
ic
e
n
v
ir
o
n
m
en
t,
Xin
g
u
an
g
et
a
l.
[
4
9
]
in
tr
o
d
u
ce
d
an
en
h
an
ce
d
v
is
u
al
SLAM
s
y
s
tem
d
esig
n
ed
f
o
r
d
y
n
am
ic
e
n
v
ir
o
n
m
e
n
ts
,
b
ased
o
n
an
im
p
r
o
v
ed
Ma
s
k
R
-
C
NN
n
eu
r
al
n
etwo
r
k
.
T
h
e
p
r
o
p
o
s
ed
SLAM
alg
o
r
ith
m
le
v
er
ag
e
s
th
e
s
em
an
tic
s
eg
m
e
n
tatio
n
ca
p
ab
ilit
ies
o
f
th
e
m
o
d
if
ied
Ma
s
k
R
-
C
NN
to
d
if
f
er
en
tiate
b
etwe
en
s
tatic
an
d
d
y
n
am
ic
p
ar
ts
o
f
th
e
s
ce
n
e
as
s
h
o
wn
i
n
Fig
u
r
e
7
.
Su
b
s
eq
u
en
tly
,
th
e
d
y
n
am
ic
f
e
atu
r
e
p
o
in
ts
ar
e
d
is
r
eg
ar
d
ed
b
y
th
e
alg
o
r
ith
m
th
at
d
etec
ts
m
o
tio
n
co
n
s
is
ten
cy
an
d
esti
m
ates
th
e
ca
m
er
a'
s
p
o
s
e
b
y
s
tatic
f
ea
tu
r
e
p
o
in
ts
in
th
e
s
tatic
r
eg
io
n
.
Fu
et
a
l.
[
5
0
]
also
p
r
o
p
o
s
ed
a
m
eth
o
d
f
o
r
d
ea
lin
g
with
d
y
n
a
m
ic
en
v
ir
o
n
m
en
ts
b
y
in
te
g
r
at
in
g
Ma
s
k
R
C
NN
with
an
att
en
tio
n
m
ec
h
an
is
m
.
T
h
e
r
esear
ch
er
s
in
teg
r
ated
th
e
co
n
v
o
lu
tio
n
al
b
lo
ck
atten
tio
n
m
o
d
u
le
(
C
B
AM
)
in
to
th
e
Ma
s
k
R
-
C
N
N
n
etwo
r
k
to
en
h
an
ce
d
y
n
a
m
ic
o
b
ject
s
e
g
m
en
tatio
n
.
T
h
ese
d
y
n
a
m
ic
o
b
ject
r
em
o
v
al
m
et
h
o
d
s
ar
e
t
h
en
co
m
b
in
ed
with
OR
B
-
SL
AM
2
[
3
3
]
.
Ho
we
v
er
,
b
o
t
h
p
r
o
p
o
s
ed
m
eth
o
d
s
r
e
m
ain
s
lo
w,
e
v
en
with
GPU
ac
ce
ler
atio
n
,
m
a
k
in
g
th
em
u
n
r
eliab
le
f
o
r
r
ea
l
-
tim
e
a
p
p
licatio
n
s
.
Fig
u
r
e
7
.
E
x
am
p
le
o
f
s
eg
m
en
t
atio
n
s
ce
n
ar
io
s
in
SLAM
s
y
s
tem
s
[
4
9
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
C
a
mera
-
b
a
s
ed
s
imu
lta
n
eo
u
s
l
o
ca
liz
a
tio
n
a
n
d
ma
p
p
in
g
…
(
A
n
a
k
A
g
u
n
g
N
g
u
r
a
h
B
a
g
u
s
Dw
ima
n
ta
r
a
)
169
Dee
p
lear
n
in
g
m
eth
o
d
s
h
av
e
also
b
ee
n
ap
p
lied
d
ir
ec
tly
as
th
e
m
eth
o
d
in
VO
as
in
[
5
1
]
,
[
5
2
]
.
W
an
g
et
a
l.
[
5
2
]
p
r
o
p
o
s
ed
a
n
o
v
el
DL
-
b
ased
m
o
n
o
cu
la
r
VO
alg
o
r
ith
m
b
y
lev
er
a
g
i
n
g
d
ee
p
r
ec
u
r
r
en
t
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
R
C
NN
s
)
[
5
3
]
,
w
h
ich
is
th
e
f
ir
s
t
en
d
-
to
-
en
d
ap
p
r
o
ac
h
o
n
th
e
m
o
n
o
cu
lar
VO
th
r
o
u
g
h
d
ee
p
n
eu
r
al
n
etwo
r
k
s
(
DNNs)
.
B
y
lev
er
ag
in
g
th
e
g
eo
m
etr
ic
f
ea
tu
r
e
r
ep
r
esen
ta
tio
n
th
at
C
NN
h
as
lear
n
ed
,
t
h
ey
s
u
g
g
ested
an
R
C
NN
ar
ch
itectu
r
e
th
at
allo
ws
th
e
DL
-
b
ased
VO
tech
n
iq
u
e
to
b
e
ap
p
lied
to
w
h
o
le
n
ew
co
n
tex
ts
.
T
h
e
KI
T
T
I
d
ataset
is
u
s
ed
to
ass
es
s
th
is
V
O's
p
er
f
o
r
m
an
c
e,
an
d
it
y
ield
s
a
tr
ajec
to
r
y
th
at
is
q
u
ite
p
r
ec
is
e
an
d
co
m
p
atib
le
with
th
e
g
r
o
u
n
d
r
ea
lity
as sh
o
wn
in
Fig
u
r
e
8
.
Fig
u
r
e
8
.
E
s
tim
ated
VO
o
n
s
eq
u
en
ce
0
5
o
f
KI
T
T
I
d
ataset
[
5
2
]
5.
F
UT
UR
E
RE
SE
A
RCH
DIS
CUSS
I
O
N
E
n
v
ir
o
n
m
en
tal
elem
en
ts
in
clu
d
in
g
illu
m
in
atio
n
,
m
o
tio
n
b
l
u
r
,
an
d
s
ce
n
e
tex
tu
r
e
ca
n
h
av
e
an
im
p
ac
t
o
n
ca
m
er
a
-
b
ased
SLAM
s
y
s
te
m
s
.
Du
e
to
th
e
ef
f
ec
t
o
f
d
y
n
a
m
ic
o
b
jects,
it
is
also
ch
allen
g
in
g
to
p
er
f
o
r
m
well
in
d
y
n
am
ic
s
itu
atio
n
s
.
Dee
p
lear
n
in
g
m
eth
o
d
s
h
av
e
b
ee
n
u
tili
ze
d
to
h
an
d
le
d
y
n
a
m
ic
o
b
j
ec
ts
,
b
u
t
r
ea
l
-
tim
e
p
er
f
o
r
m
an
ce
is
h
ar
d
to
ac
h
iev
e.
Ad
d
itio
n
ally
,
to
in
cr
ea
s
e
p
r
o
ce
s
s
in
g
s
p
ee
d
with
o
u
t
co
m
p
r
o
m
is
in
g
ac
c
u
r
ac
y
,
o
p
tim
izatio
n
s
tr
ateg
ies
lik
e
GPU
ac
ce
ler
atio
n
an
d
s
p
ar
s
e
r
ep
r
esen
tatio
n
s
ar
e
u
s
ed
.
Desp
ite
th
ese
d
ev
elo
p
m
e
n
ts
,
th
er
e
is
s
till
a
c
h
allen
g
e
in
s
tr
ik
in
g
a
b
ala
n
ce
b
etwe
en
r
o
b
u
s
t
h
an
d
lin
g
o
f
d
y
n
am
ic
asp
ec
ts
an
d
r
ea
l
-
tim
e
s
p
ee
d
,
w
h
ich
m
o
tiv
ates
co
n
tin
u
ed
s
tu
d
y
in
t
o
alg
o
r
ith
m
ef
f
icien
cy
a
n
d
a
d
ap
tat
io
n
to
a
v
ar
iety
o
f
r
ea
l
-
wo
r
ld
s
itu
atio
n
s
6.
CO
NCLU
SI
O
N
P
a
s
t
r
e
s
e
a
r
ch
o
n
s
im
u
l
t
an
eo
u
s
lo
c
a
l
i
z
a
t
io
n
a
n
d
m
a
p
p
in
g
(
S
L
A
M
)
h
a
s
a
c
h
i
ev
ed
s
i
g
n
i
f
i
ca
n
t
a
d
v
an
c
em
e
n
t
s
th
r
o
u
g
h
tr
a
d
i
ti
o
n
a
l
te
c
h
n
i
q
u
e
s
l
i
k
e
i
n
d
i
r
ec
t
a
n
d
d
ir
e
c
t
m
e
t
h
o
d
s
,
wh
ic
h
l
ev
e
r
ag
e
r
o
b
u
s
t
g
e
o
m
e
t
r
i
c
an
d
p
h
o
t
o
m
e
t
r
i
c
d
a
t
a
f
o
r
a
cc
u
r
a
t
e
l
o
c
a
l
i
z
a
t
io
n
a
n
d
m
ap
p
in
g
.
I
n
n
o
v
a
t
i
o
n
s
i
n
l
o
o
p
c
l
o
s
u
r
e
d
e
t
e
c
t
io
n
an
d
th
e
u
s
e
o
f
v
a
r
io
u
s
c
a
m
e
r
a
s
y
s
t
em
s
,
s
u
c
h
a
s
m
o
n
o
cu
l
a
r
,
s
t
er
e
o
,
a
n
d
R
G
B
-
D
,
h
a
v
e
a
d
d
r
e
s
s
e
d
c
h
a
l
l
en
g
e
s
l
i
k
e
s
c
a
le
a
m
b
ig
u
i
ty
,
d
ep
t
h
e
s
t
i
m
a
t
i
o
n
,
a
n
d
s
c
en
e
co
m
p
le
x
i
ty
,
w
i
t
h
t
r
a
d
e
-
o
f
f
s
b
e
t
w
ee
n
s
i
m
p
l
i
c
i
t
y
a
n
d
s
p
a
t
ia
l
i
n
f
o
r
m
a
t
i
o
n
r
i
ch
n
e
s
s
.
T
h
e
i
n
t
eg
r
at
i
o
n
o
f
d
e
ep
l
e
ar
n
in
g
h
a
s
f
u
r
t
h
er
r
ev
o
lu
t
i
o
n
i
z
ed
S
L
A
M
b
y
en
h
an
c
i
n
g
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
,
e
n
v
i
r
o
n
m
e
n
t
u
n
d
e
r
s
t
an
d
in
g
,
a
n
d
d
y
n
a
m
i
c
o
b
j
e
c
t
s
eg
m
en
t
a
t
i
o
n
,
e
n
a
b
l
in
g
i
m
p
r
o
v
ed
r
o
b
u
s
t
n
es
s
a
n
d
ad
a
p
ta
b
i
l
i
ty
i
n
c
o
m
p
l
e
x
s
c
en
ar
i
o
s
.
De
s
p
i
t
e
th
es
e
a
d
v
an
c
em
e
n
t
s
,
a
c
h
i
ev
i
n
g
r
e
a
l
-
t
im
e
p
er
f
o
r
m
a
n
c
e
a
n
d
s
c
a
l
a
b
i
l
i
ty
i
n
d
iv
e
r
s
e
,
c
o
m
p
l
ex
e
n
v
i
r
o
n
m
e
n
t
s
r
em
a
i
n
s
a
m
a
jo
r
c
h
a
l
l
en
g
e.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
wo
r
k
is
s
u
p
p
o
r
ted
b
y
Un
iv
er
s
itas
Gad
jah
Ma
d
a
u
n
d
er
th
e
f
u
n
d
in
g
o
f
th
e
Fin
al
Pro
ject
R
ec
o
g
n
itio
n
Pro
g
r
am
(
R
T
A,
c
o
n
tr
ac
t n
u
m
b
er
:
5
2
8
6
/UN1
.
P1
/PT.
0
1
.
0
3
/2
0
2
4
)
f
o
r
th
e
y
e
ar
2
0
2
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
14
,
No
.
2
,
J
u
n
e
20
2
5
:
1
62
-
1
72
170
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
An
ak
Ag
u
n
g
Ng
u
r
a
h
B
ag
u
s
Dwim
an
tar
a
✓
✓
✓
✓
✓
Osk
ar
Nata
n
✓
✓
✓
✓
✓
✓
✓
✓
✓
No
v
elio
Pu
tr
a
I
n
d
ar
to
✓
✓
✓
✓
An
d
i D
h
ar
m
awa
n
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
Data
a
v
aila
b
i
lit
y
is
n
o
t
a
p
p
li
ca
b
le
t
o
t
h
is
p
ap
er
as
n
o
n
e
w
d
a
t
a
w
er
e
cr
ea
te
d
o
r
an
al
y
z
e
d
i
n
t
h
is
s
t
u
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
B
.
G
a
o
,
H
.
La
n
g
,
a
n
d
J
.
R
e
n
,
“
S
t
e
r
e
o
v
i
s
u
a
l
S
LA
M
f
o
r
a
u
t
o
n
o
m
o
u
s
v
e
h
i
c
l
e
s:
A
r
e
v
i
e
w
,
”
i
n
C
o
n
f
e
r
e
n
c
e
Pro
c
e
e
d
i
n
g
s
-
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
y
st
e
m
s,
M
a
n
a
n
d
C
y
b
e
r
n
e
t
i
c
s
,
O
c
t
.
2
0
2
0
,
v
o
l
.
2
0
2
0
-
O
c
t
o
b
,
p
p
.
1
3
1
6
–
1
3
2
2
.
d
o
i
:
1
0
.
1
1
0
9
/
S
M
C
4
2
9
7
5
.
2
0
2
0
.
9
2
8
3
1
6
1
.
[
2
]
C
.
P
a
n
g
,
L.
Z
h
o
u
,
a
n
d
X
.
H
u
a
n
g
,
“
A
l
o
w
-
c
o
s
t
3
D
S
LA
M
s
y
s
t
e
m
i
n
t
e
g
r
a
t
i
o
n
o
f
a
u
t
o
n
o
m
o
u
s
e
x
p
l
o
r
a
t
i
o
n
b
a
sed
o
n
F
a
st
-
I
C
P
e
n
h
a
n
c
e
d
Li
D
A
R
-
i
n
e
r
t
i
a
l
o
d
o
me
t
r
y
,
”
Re
m
o
t
e
S
e
n
s
i
n
g
,
v
o
l
.
1
6
,
n
o
.
1
1
,
p
.
1
9
7
9
,
M
a
y
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s1
6
1
1
1
9
7
9
.
[
3
]
X
.
H
a
n
,
Y
.
Ta
o
,
Z.
L
i
,
R
.
C
e
n
,
a
n
d
F
.
X
u
e
,
“
S
u
p
e
r
P
o
i
n
t
V
O
:
A
l
i
g
h
t
w
e
i
g
h
t
v
i
s
u
a
l
o
d
o
m
e
t
r
y
b
a
se
d
o
n
C
N
N
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
u
t
o
m
a
t
i
o
n
,
C
o
n
t
r
o
l
a
n
d
Ro
b
o
t
i
c
s
En
g
i
n
e
e
r
i
n
g
,
C
AC
RE
2
0
2
0
,
S
e
p
.
2
0
2
0
,
p
p
.
6
8
5
–
6
9
1
.
d
o
i
:
1
0
.
1
1
0
9
/
C
A
C
R
E
5
0
1
3
8
.
2
0
2
0
.
9
2
3
0
3
4
8
.
[
4
]
J.
S
h
i
a
n
d
C
.
To
masi
,
“
G
o
o
d
f
e
a
t
u
r
e
s
t
o
t
r
a
c
k
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
1
9
9
4
,
p
p
.
5
9
3
–
6
0
0
.
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
.
1
9
9
4
.
3
2
3
7
9
4
.
[
5
]
R
.
M
u
r
-
A
r
t
a
l
,
J.
M
.
M
.
M
o
n
t
i
e
l
,
a
n
d
J
.
D
.
Ta
r
d
o
s,
“
O
R
B
-
S
LA
M
:
A
v
e
r
sat
i
l
e
a
n
d
a
c
c
u
r
a
t
e
m
o
n
o
c
u
l
a
r
S
LA
M
s
y
s
t
e
m
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
R
o
b
o
t
i
c
s
,
v
o
l
.
3
1
,
n
o
.
5
,
p
p
.
1
1
4
7
–
1
1
6
3
,
O
c
t
.
2
0
1
5
,
d
o
i
:
1
0
.
1
1
0
9
/
T
R
O
.
2
0
1
5
.
2
4
6
3
6
7
1
.
[
6
]
J.
E
n
g
e
l
,
T.
S
c
h
ö
p
s,
a
n
d
D
.
C
r
e
m
e
r
s,
“
LSD
-
S
LA
M
:
La
r
g
e
-
sc
a
l
e
d
i
r
e
c
t
m
o
n
o
c
u
l
a
r
S
LA
M
,
”
i
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
su
b
ser
i
e
s
L
e
c
t
u
re
N
o
t
e
s
i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
re
N
o
t
e
s
i
n
B
i
o
i
n
f
o
rm
a
t
i
c
s)
,
v
o
l
.
8
6
9
0
LN
C
S
,
n
o
.
P
A
R
T
2
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
sh
i
n
g
,
2
0
1
4
,
p
p
.
8
3
4
–
8
4
9
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
1
0
6
0
5
-
2
_
5
4
.
[
7
]
A
.
J.
D
a
v
i
s
o
n
,
I
.
D
.
R
e
i
d
,
N
.
D
.
M
o
l
t
o
n
,
a
n
d
O
.
S
t
a
ss
e
,
“
M
o
n
o
S
LA
M
:
r
e
a
l
-
t
i
me
si
n
g
l
e
c
a
mera
S
LA
M
,
”
I
EEE
T
ra
n
sa
c
t
i
o
n
s
o
n
Pa
t
t
e
r
n
An
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
2
9
,
n
o
.
6
,
p
p
.
1
0
5
2
–
1
0
6
7
,
J
u
n
.
2
0
0
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
0
7
.
1
0
4
9
.
[
8
]
J.
En
g
e
l
,
J
.
S
t
ü
c
k
l
e
r
,
a
n
d
D
.
C
r
e
m
e
r
s,
“
La
r
g
e
-
s
c
a
l
e
d
i
r
e
c
t
S
LA
M
w
i
t
h
s
t
e
r
e
o
c
a
m
e
r
a
s,
”
i
n
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
Ro
b
o
t
s
a
n
d
S
y
st
e
m
s
,
S
e
p
.
2
0
1
5
,
v
o
l
.
2
0
1
5
-
D
e
c
e
mb
e
r
,
p
p
.
1
9
3
5
–
1
9
4
2
.
d
o
i
:
1
0
.
1
1
0
9
/
I
R
O
S
.
2
0
1
5
.
7
3
5
3
6
3
1
.
[
9
]
C
.
K
e
r
l
,
J.
S
t
u
r
m,
a
n
d
D
.
C
r
e
mers
,
“
R
o
b
u
st
o
d
o
me
t
r
y
e
st
i
ma
t
i
o
n
f
o
r
R
G
B
-
D
c
a
m
e
r
a
s,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
R
o
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
3
,
p
p
.
3
7
4
8
–
3
7
5
4
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
3
.
6
6
3
1
1
0
4
.
[
1
0
]
A
.
G
e
i
g
e
r
,
P
.
Le
n
z
,
a
n
d
R
.
U
r
t
a
su
n
,
“
A
r
e
w
e
r
e
a
d
y
f
o
r
a
u
t
o
n
o
m
o
u
s
d
r
i
v
i
n
g
?
t
h
e
K
I
TTI
v
i
si
o
n
b
e
n
c
h
mar
k
s
u
i
t
e
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
s
i
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
1
2
,
p
p
.
3
3
5
4
–
3
3
6
1
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
2
.
6
2
4
8
0
7
4
.
[
1
1
]
M
.
R
a
mez
a
n
i
,
Y
.
W
a
n
g
,
M
.
C
a
mu
r
r
i
,
D
.
W
i
st
h
,
M
.
M
a
t
t
a
ma
l
a
,
a
n
d
M
.
F
a
l
l
o
n
,
“
T
h
e
n
e
w
e
r
c
o
l
l
e
g
e
d
a
t
a
s
e
t
:
H
a
n
d
h
e
l
d
Li
D
A
R
,
i
n
e
r
t
i
a
l
a
n
d
v
i
si
o
n
w
i
t
h
g
r
o
u
n
d
t
r
u
t
h
,
”
i
n
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
R
o
b
o
t
s
a
n
d
S
y
st
e
m
s
,
O
c
t
.
2
0
2
0
,
p
p
.
4
3
5
3
–
4
3
6
0
.
d
o
i
:
1
0
.
1
1
0
9
/
I
R
O
S
4
5
7
4
3
.
2
0
2
0
.
9
3
4
0
8
4
9
.
[
1
2
]
J.
En
g
e
l
,
V
.
U
se
n
k
o
,
a
n
d
D
.
C
r
e
mers
,
“
A
p
h
o
t
o
me
t
r
i
c
a
l
l
y
c
a
l
i
b
r
a
t
e
d
b
e
n
c
h
mar
k
f
o
r
mo
n
o
c
u
l
a
r
v
i
su
a
l
o
d
o
me
t
r
y
,
”
a
rXi
v
:
1
6
0
7
.
0
2
5
5
5
,
Ju
l
.
2
0
1
6
.
[
1
3
]
J.
S
t
u
r
m,
N
.
E
n
g
e
l
h
a
r
d
,
F
.
En
d
r
e
s
,
W
.
B
u
r
g
a
r
d
,
a
n
d
D
.
C
r
e
mers
,
“
A
b
e
n
c
h
mark
f
o
r
t
h
e
e
v
a
l
u
a
t
i
o
n
o
f
R
G
B
-
D
S
LA
M
sy
s
t
e
ms
,
”
i
n
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
Ro
b
o
t
s
a
n
d
S
y
st
e
m
s
,
O
c
t
.
2
0
1
2
,
p
p
.
5
7
3
–
5
8
0
.
d
o
i
:
1
0
.
1
1
0
9
/
I
R
O
S
.
2
0
1
2
.
6
3
8
5
7
7
3
.
[
1
4
]
M
.
B
u
r
r
i
e
t
a
l
.
,
“
Th
e
E
u
R
o
C
m
i
c
r
o
a
e
r
i
a
l
v
e
h
i
c
l
e
d
a
t
a
se
t
s,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
R
o
b
o
t
i
c
s
Re
se
a
r
c
h
,
v
o
l
.
3
5
,
n
o
.
1
0
,
p
p
.
1
1
5
7
–
1
1
6
3
,
J
a
n
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
7
7
/
0
2
7
8
3
6
4
9
1
5
6
2
0
0
3
3
.
[
1
5
]
J.
E
n
g
e
l
,
V
.
K
o
l
t
u
n
,
a
n
d
D
.
C
r
e
mers
,
“
D
i
r
e
c
t
s
p
a
r
se
o
d
o
me
t
r
y
,
”
I
EE
E
T
r
a
n
sa
c
t
i
o
n
s
o
n
Pa
t
t
e
rn
An
a
l
y
s
i
s a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
0
,
n
o
.
3
,
p
p
.
6
1
1
–
6
2
5
,
M
a
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
7
.
2
6
5
8
5
7
7
.
[
1
6
]
T.
W
h
e
l
a
n
,
R
.
F
.
S
a
l
a
s
-
M
o
r
e
n
o
,
B
.
G
l
o
c
k
e
r
,
A
.
J
.
D
a
v
i
s
o
n
,
a
n
d
S
.
Le
u
t
e
n
e
g
g
e
r
,
“
E
l
a
s
t
i
c
F
u
s
i
o
n
:
R
e
a
l
-
t
i
me
d
e
n
se
S
LA
M
a
n
d
l
i
g
h
t
so
u
r
c
e
e
s
t
i
m
a
t
i
o
n
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
R
o
b
o
t
i
c
s
Re
s
e
a
r
c
h
,
v
o
l
.
3
5
,
n
o
.
1
4
,
p
p
.
1
6
9
7
–
1
7
1
6
,
S
e
p
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
7
7
/
0
2
7
8
3
6
4
9
1
6
6
6
9
2
3
7
.
[
1
7
]
B
.
G
l
o
c
k
e
r
,
J.
S
h
o
t
t
o
n
,
A
.
C
r
i
mi
n
i
s
i
,
a
n
d
S
.
I
z
a
d
i
,
“
R
e
a
l
-
t
i
m
e
R
G
B
-
D
c
a
mera
r
e
l
o
c
a
l
i
z
a
t
i
o
n
v
i
a
r
a
n
d
o
m
i
z
e
d
f
e
r
n
s
f
o
r
k
e
y
f
r
a
me
e
n
c
o
d
i
n
g
,
”
I
E
EE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
Vi
su
a
l
i
z
a
t
i
o
n
a
n
d
C
o
m
p
u
t
e
r
G
ra
p
h
i
c
s
,
v
o
l
.
2
1
,
n
o
.
5
,
p
p
.
5
7
1
–
5
8
3
,
M
a
y
2
0
1
5
,
d
o
i
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
C
a
mera
-
b
a
s
ed
s
imu
lta
n
eo
u
s
l
o
ca
liz
a
tio
n
a
n
d
ma
p
p
in
g
…
(
A
n
a
k
A
g
u
n
g
N
g
u
r
a
h
B
a
g
u
s
Dw
ima
n
ta
r
a
)
171
1
0
.
1
1
0
9
/
TV
C
G
.
2
0
1
4
.
2
3
6
0
4
0
3
.
[
1
8
]
J.
En
g
e
l
,
J
.
S
t
u
r
m,
a
n
d
D
.
C
r
e
mers
,
“
S
e
m
i
-
d
e
n
s
e
v
i
su
a
l
o
d
o
m
e
t
r
y
f
o
r
a
mo
n
o
c
u
l
a
r
c
a
m
e
r
a
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r V
i
s
i
o
n
,
D
e
c
.
2
0
1
3
,
p
p
.
1
4
4
9
–
1
4
5
6
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
1
3
.
1
8
3
.
[
1
9
]
A
.
G
l
o
v
e
r
,
W
.
M
a
d
d
e
r
n
,
M
.
W
a
r
r
e
n
,
S
.
R
e
i
d
,
M
.
M
i
l
f
o
r
d
,
a
n
d
G
.
W
y
e
t
h
,
“
O
p
e
n
F
A
B
M
A
P
:
A
n
o
p
e
n
so
u
r
c
e
t
o
o
l
b
o
x
f
o
r
a
p
p
e
a
r
a
n
c
e
-
b
a
s
e
d
l
o
o
p
c
l
o
su
r
e
d
e
t
e
c
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ro
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
2
,
p
p
.
4
7
3
0
–
4
7
3
5
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
2
.
6
2
2
4
8
4
3
.
[
2
0
]
X
.
G
a
o
,
R
.
W
a
n
g
,
N
.
D
e
mm
e
l
,
a
n
d
D
.
C
r
e
mers
,
“
LD
S
O
:
D
i
r
e
c
t
sp
a
r
se
o
d
o
me
t
r
y
w
i
t
h
l
o
o
p
c
l
o
su
r
e
,
”
i
n
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
R
o
b
o
t
s
a
n
d
S
y
s
t
e
m
s
,
O
c
t
.
2
0
1
8
,
p
p
.
2
1
9
8
–
2
2
0
4
.
d
o
i
:
1
0
.
1
1
0
9
/
I
R
O
S
.
2
0
1
8
.
8
5
9
3
3
7
6
.
[
2
1
]
D
.
G
á
l
v
e
z
-
Ló
p
e
z
a
n
d
J.
D
.
T
a
r
d
ó
s,
“
B
a
g
s
o
f
b
i
n
a
r
y
w
o
r
d
s
f
o
r
f
a
s
t
p
l
a
c
e
r
e
c
o
g
n
i
t
i
o
n
i
n
i
ma
g
e
se
q
u
e
n
c
e
s
,
”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
Ro
b
o
t
i
c
s
,
v
o
l
.
2
8
,
n
o
.
5
,
p
p
.
1
1
8
8
–
1
1
9
7
,
O
c
t
.
2
0
1
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TR
O
.
2
0
1
2
.
2
1
9
7
1
5
8
.
[
2
2
]
H
.
C
h
o
,
E.
K
.
K
i
m,
a
n
d
S
.
K
i
m,
“
I
n
d
o
o
r
S
LA
M
a
p
p
l
i
c
a
t
i
o
n
u
s
i
n
g
g
e
o
m
e
t
r
i
c
a
n
d
I
C
P
mat
c
h
i
n
g
me
t
h
o
d
s
b
a
se
d
o
n
l
i
n
e
f
e
a
t
u
r
e
s
,
”
Ro
b
o
t
i
c
s
a
n
d
Au
t
o
n
o
m
o
u
s
S
y
s
t
e
m
s
,
v
o
l
.
1
0
0
,
p
p
.
2
0
6
–
2
2
4
,
F
e
b
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
r
o
b
o
t
.
2
0
1
7
.
1
1
.
0
1
1
.
[
2
3
]
A
.
K
a
s
y
a
n
o
v
,
F
.
En
g
e
l
m
a
n
n
,
J.
S
t
u
c
k
l
e
r
,
a
n
d
B
.
Le
i
b
e
,
“
K
e
y
f
r
a
m
e
-
b
a
s
e
d
v
i
su
a
l
-
i
n
e
r
t
i
a
l
o
n
l
i
n
e
S
LA
M
w
i
t
h
r
e
l
o
c
a
l
i
z
a
t
i
o
n
,
”
i
n
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
Ro
b
o
t
s
a
n
d
S
y
s
t
e
m
s
,
S
e
p
.
2
0
1
7
,
v
o
l
.
2
0
1
7
-
S
e
p
t
e
m
b
e
r
,
p
p
.
6
6
6
2
–
6
6
6
9
.
d
o
i
:
1
0
.
1
1
0
9
/
I
R
O
S
.
2
0
1
7
.
8
2
0
6
5
8
1
.
[
2
4
]
H
.
L
i
m,
J.
L
i
m,
a
n
d
H
.
J.
K
i
m,
“
R
e
a
l
-
t
i
m
e
6
-
D
O
F
m
o
n
o
c
u
l
a
r
v
i
su
a
l
S
LA
M
i
n
a
l
a
r
g
e
-
s
c
a
l
e
e
n
v
i
r
o
n
me
n
t
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
R
o
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
4
,
p
p
.
1
5
3
2
–
1
5
3
9
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
4
.
6
9
0
7
0
5
5
.
[
2
5
]
D
.
Li
e
t
a
l
.
,
“
A
S
LA
M
s
y
st
e
m
b
a
s
e
d
o
n
R
G
B
D
i
ma
g
e
a
n
d
p
o
i
n
t
-
l
i
n
e
f
e
a
t
u
r
e
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
9
0
1
2
–
9
0
2
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
4
9
4
6
7
.
[
2
6
]
H
.
B
a
y
,
A
.
Ess,
T.
T
u
y
t
e
l
a
a
r
s,
a
n
d
L.
V
a
n
G
o
o
l
,
“
S
p
e
e
d
e
d
-
u
p
r
o
b
u
st
f
e
a
t
u
r
e
s
(
S
U
R
F
)
,
”
C
o
m
p
u
t
e
r
V
i
s
i
o
n
a
n
d
I
m
a
g
e
U
n
d
e
rst
a
n
d
i
n
g
,
v
o
l
.
1
1
0
,
n
o
.
3
,
p
p
.
3
4
6
–
3
5
9
,
J
u
n
.
2
0
0
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
v
i
u
.
2
0
0
7
.
0
9
.
0
1
4
.
[
2
7
]
E.
R
u
b
l
e
e
,
V
.
R
a
b
a
u
d
,
K
.
K
o
n
o
l
i
g
e
,
a
n
d
G
.
B
r
a
d
sk
i
,
“
O
R
B
:
A
n
e
f
f
i
c
i
e
n
t
a
l
t
e
r
n
a
t
i
v
e
t
o
S
I
F
T
o
r
S
U
R
F
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
,
2
0
1
1
,
p
p
.
2
5
6
4
–
2
5
7
1
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
1
1
.
6
1
2
6
5
4
4
.
[
2
8
]
F
.
N
o
b
i
s,
O
.
P
a
p
a
n
i
k
o
l
a
o
u
,
J
.
B
e
t
z
,
a
n
d
M
.
Li
e
n
k
a
mp
,
“
P
e
r
si
st
e
n
t
m
a
p
s
a
v
i
n
g
f
o
r
v
i
s
u
a
l
l
o
c
a
l
i
z
a
t
i
o
n
f
o
r
a
u
t
o
n
o
mo
u
s
v
e
h
i
c
l
e
s
:
A
n
O
R
B
-
S
LA
M
2
e
x
t
e
n
si
o
n
,
”
i
n
2
0
2
0
1
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ec
o
l
o
g
i
c
a
l
Ve
h
i
c
l
e
s
a
n
d
R
e
n
e
w
a
b
l
e
E
n
e
r
g
i
e
s,
EV
ER
2
0
2
0
,
S
e
p
.
2
0
2
0
,
p
p
.
1
–
9
.
d
o
i
:
1
0
.
1
1
0
9
/
EV
E
R
4
8
7
7
6
.
2
0
2
0
.
9
2
4
3
0
9
4
.
[
2
9
]
T.
Q
i
n
,
P
.
Li
,
a
n
d
S
.
S
h
e
n
,
“
V
I
N
S
-
M
o
n
o
:
A
r
o
b
u
st
a
n
d
v
e
r
sat
i
l
e
mo
n
o
c
u
l
a
r
v
i
su
a
l
-
i
n
e
r
t
i
a
l
s
t
a
t
e
e
s
t
i
m
a
t
o
r
,
”
I
E
E
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
Ro
b
o
t
i
c
s
,
v
o
l
.
3
4
,
n
o
.
4
,
p
p
.
1
0
0
4
–
1
0
2
0
,
A
u
g
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
T
R
O
.
2
0
1
8
.
2
8
5
3
7
2
9
.
[
3
0
]
J.
M
.
M
a
r
t
í
n
e
z
-
O
t
z
e
t
a
,
I
.
R
o
d
r
í
g
u
e
z
-
M
o
r
e
n
o
,
I
.
M
e
n
d
i
a
l
d
u
a
,
a
n
d
B
.
S
i
e
r
r
a
,
“
R
A
N
S
A
C
f
o
r
r
o
b
o
t
i
c
a
p
p
l
i
c
a
t
i
o
n
s
:
A
su
r
v
e
y
,
”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
1
,
p
.
3
2
7
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
3
0
1
0
3
2
7
.
[
3
1
]
J.
S
t
a
l
b
a
u
m
a
n
d
J.
B
.
S
o
n
g
,
“
K
e
y
f
r
a
me
a
n
d
i
n
l
i
e
r
s
e
l
e
c
t
i
o
n
f
o
r
v
i
su
a
l
S
LA
M
,
”
i
n
2
0
1
3
1
0
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
U
b
i
q
u
i
t
o
u
s R
o
b
o
t
s
a
n
d
Am
b
i
e
n
t
I
n
t
e
l
l
i
g
e
n
c
e
,
U
RAI
2
0
1
3
,
O
c
t
.
2
0
1
3
,
p
p
.
3
9
1
–
3
9
6
.
d
o
i
:
1
0
.
1
1
0
9
/
U
R
A
I
.
2
0
1
3
.
6
6
7
7
2
9
5
.
[
3
2
]
R
.
K
ü
mm
e
r
l
e
,
G
.
G
r
i
s
e
t
t
i
,
H
.
S
t
r
a
sd
a
t
,
K
.
K
o
n
o
l
i
g
e
,
a
n
d
W
.
B
u
r
g
a
r
d
,
“
G
2
o
:
A
g
e
n
e
r
a
l
f
r
a
m
e
w
o
r
k
f
o
r
g
r
a
p
h
o
p
t
i
m
i
z
a
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
R
o
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
1
,
p
p
.
3
6
0
7
–
3
6
1
3
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
1
.
5
9
7
9
9
4
9
.
[
3
3
]
R
.
M
u
r
-
A
r
t
a
l
a
n
d
J.
D
.
T
a
r
d
o
s,
“
O
R
B
-
S
LA
M
2
:
A
n
o
p
e
n
-
s
o
u
r
c
e
S
LA
M
sy
st
e
m
f
o
r
m
o
n
o
c
u
l
a
r
,
s
t
e
r
e
o
,
a
n
d
R
G
B
-
D
c
a
mer
a
s,”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
R
o
b
o
t
i
c
s
,
v
o
l
.
3
3
,
n
o
.
5
,
p
p
.
1
2
5
5
–
1
2
6
2
,
O
c
t
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TR
O
.
2
0
1
7
.
2
7
0
5
1
0
3
.
[
3
4
]
A
.
H
a
n
d
a
,
T.
W
h
e
l
a
n
,
J.
M
c
D
o
n
a
l
d
,
a
n
d
A
.
J.
D
a
v
i
s
o
n
,
“
A
b
e
n
c
h
m
a
r
k
f
o
r
R
G
B
-
D
v
i
s
u
a
l
o
d
o
met
r
y
,
3
D
r
e
c
o
n
s
t
r
u
c
t
i
o
n
a
n
d
S
LA
M
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
R
o
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
4
,
p
p
.
1
5
2
4
–
1
5
3
1
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
4
.
6
9
0
7
0
5
4
.
[
3
5
]
C
.
F
o
r
st
e
r
,
M
.
P
i
z
z
o
l
i
,
a
n
d
D
.
S
c
a
r
a
mu
z
z
a
,
“
S
V
O
:
F
a
st
sem
i
-
d
i
r
e
c
t
mo
n
o
c
u
l
a
r
v
i
su
a
l
o
d
o
me
t
r
y
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
R
o
b
o
t
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
,
M
a
y
2
0
1
4
,
p
p
.
1
5
–
2
2
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
A
.
2
0
1
4
.
6
9
0
6
5
8
4
.
[
3
6
]
X
.
Zh
a
o
,
L
.
L
i
u
,
R
.
Zh
e
n
g
,
W
.
Y
e
,
a
n
d
Y
.
Li
u
,
“
A
r
o
b
u
st
st
e
r
e
o
f
e
a
t
u
r
e
-
a
i
d
e
d
sem
i
-
d
i
r
e
c
t
S
LA
M
s
y
st
e
m
,
”
R
o
b
o
t
i
c
s
a
n
d
Au
t
o
n
o
m
o
u
s
S
y
s
t
e
m
s
,
v
o
l
.
1
3
2
,
p
.
1
0
3
5
9
7
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
r
o
b
o
t
.
2
0
2
0
.
1
0
3
5
9
7
.
[
3
7
]
X
.
D
o
n
g
,
L
.
C
h
e
n
g
,
H
.
P
e
n
g
,
a
n
d
T.
Li
,
“
F
S
D
-
S
LA
M
:
a
f
a
s
t
sem
i
-
d
i
r
e
c
t
S
LA
M
a
l
g
o
r
i
t
h
m
,
”
C
o
m
p
l
e
x
a
n
d
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
8
,
n
o
.
3
,
p
p
.
1
8
2
3
–
1
8
3
4
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s4
0
7
4
7
-
0
2
1
-
0
0
3
2
3
-
y.
[
3
8
]
S
.
W
e
i
ss
e
t
a
l
.
,
“
M
o
n
o
c
u
l
a
r
v
i
s
i
o
n
f
o
r
l
o
n
g
-
t
e
r
m
mi
c
r
o
a
e
r
i
a
l
v
e
h
i
c
l
e
s
t
a
t
e
e
st
i
m
a
t
i
o
n
:
A
c
o
m
p
e
n
d
i
u
m,”
J
o
u
r
n
a
l
o
f
Fi
e
l
d
Ro
b
o
t
i
c
s
,
v
o
l
.
3
0
,
n
o
.
5
,
p
p
.
8
0
3
–
8
3
1
,
A
u
g
.
2
0
1
3
,
d
o
i
:
1
0
.
1
0
0
2
/
r
o
b
.
2
1
4
6
6
.
[
3
9
]
G
.
K
l
e
i
n
a
n
d
D
.
M
u
r
r
a
y
,
“
P
a
r
a
l
l
e
l
t
r
a
c
k
i
n
g
a
n
d
ma
p
p
i
n
g
f
o
r
sma
l
l
A
R
w
o
r
k
s
p
a
c
e
s,
”
i
n
2
0
0
7
6
t
h
I
E
EE
a
n
d
A
C
M
I
n
t
e
rn
a
t
i
o
n
a
l
S
y
m
p
o
s
i
u
m
o
n
Mi
x
e
d
a
n
d
Au
g
m
e
n
t
e
d
Re
a
l
i
t
y
,
I
S
MA
R
,
N
o
v
.
2
0
0
7
,
p
p
.
2
2
5
–
2
3
4
.
d
o
i
:
1
0
.
1
1
0
9
/
I
S
M
A
R
.
2
0
0
7
.
4
5
3
8
8
5
2
.
[
4
0
]
R
.
W
a
n
g
,
M
.
S
c
h
w
o
r
e
r
,
a
n
d
D
.
C
r
e
mers
,
“
S
t
e
r
e
o
D
S
O
:
La
r
g
e
-
sc
a
l
e
d
i
r
e
c
t
sp
a
r
se
v
i
s
u
a
l
o
d
o
m
e
t
r
y
w
i
t
h
st
e
r
e
o
c
a
m
e
r
a
s,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
,
O
c
t
.
2
0
1
7
,
v
o
l
.
2
0
1
7
-
O
c
t
o
b
,
p
p
.
3
9
2
3
–
3
9
3
1
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
1
7
.
4
2
1
.
[
4
1
]
N
.
K
r
o
m
b
a
c
h
,
D
.
D
r
o
e
sc
h
e
l
,
a
n
d
S
.
B
e
h
n
k
e
,
“
C
o
m
b
i
n
i
n
g
f
e
a
t
u
r
e
-
b
a
se
d
a
n
d
d
i
r
e
c
t
me
t
h
o
d
s
f
o
r
sem
i
-
d
e
n
s
e
r
e
a
l
-
t
i
me
st
e
r
e
o
v
i
s
u
a
l
o
d
o
m
e
t
r
y
,
”
i
n
Ad
v
a
n
c
e
s
i
n
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
5
3
1
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
s
h
i
n
g
,
2
0
1
7
,
p
p
.
8
5
5
–
8
6
8
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
319
-
4
8
0
3
6
-
7
_
6
2
.
[
4
2
]
M
.
C
o
r
d
t
s
e
t
a
l
.
,
“
T
h
e
c
i
t
y
sc
a
p
e
s
d
a
t
a
set
f
o
r
se
ma
n
t
i
c
u
r
b
a
n
sc
e
n
e
u
n
d
e
r
st
a
n
d
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m
,
p
p
.
3
2
1
3
–
3
2
2
3
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
3
5
0
.
[
4
3
]
B
.
R
.
S
o
l
u
n
k
e
a
n
d
S
.
R
.
G
e
n
g
a
j
e
,
“
A
r
e
v
i
e
w
o
n
t
r
a
d
i
t
i
o
n
a
l
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
b
a
se
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
me
t
h
o
d
s,”
M
a
r
.
2
0
2
3
.
d
o
i
:
1
0
.
1
1
0
9
/
ES
C
I
5
6
8
7
2
.
2
0
2
3
.
1
0
0
9
9
6
3
9
.
[
4
4
]
D
.
D
e
t
o
n
e
,
T.
M
a
l
i
si
e
w
i
c
z
,
a
n
d
A
.
R
a
b
i
n
o
v
i
c
h
,
“
S
u
p
e
r
P
o
i
n
t
:
S
e
l
f
-
su
p
e
r
v
i
s
e
d
i
n
t
e
r
e
st
p
o
i
n
t
d
e
t
e
c
t
i
o
n
a
n
d
d
e
s
c
r
i
p
t
i
o
n
,
”
i
n
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
W
o
r
k
sh
o
p
s
,
Ju
n
.
2
0
1
8
,
v
o
l
.
2
0
1
8
-
J
u
n
e
,
p
p
.
3
3
7
–
3
4
9
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
8
.
0
0
0
6
0
.
[
4
5
]
C
.
H
a
messe
,
M
.
V
l
a
m
i
n
c
k
,
H
.
L
u
o
n
g
,
a
n
d
R
.
H
a
e
l
t
e
r
m
a
n
,
“
P
r
a
c
t
i
c
a
l
d
e
e
p
f
e
a
t
u
r
e
-
b
a
s
e
d
v
i
s
u
a
l
-
i
n
e
r
t
i
a
l
o
d
o
me
t
r
y
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Pa
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
A
p
p
l
i
c
a
t
i
o
n
s
a
n
d
M
e
t
h
o
d
s
,
2
0
2
4
,
v
o
l
.
1
,
p
p
.
2
4
0
–
2
4
7
.
d
o
i
:
1
0
.
5
2
2
0
/
0
0
1
2
3
2
0
2
0
0
0
0
3
6
5
4
.
[
4
6
]
P
.
Li
n
d
e
n
b
e
r
g
e
r
,
P
.
E.
S
a
r
l
i
n
,
a
n
d
M
.
P
o
l
l
e
f
e
y
s
,
“
Li
g
h
t
G
l
u
e
:
L
o
c
a
l
f
e
a
t
u
r
e
mat
c
h
i
n
g
a
t
l
i
g
h
t
s
p
e
e
d
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r V
i
s
i
o
n
,
O
c
t
.
2
0
2
3
,
p
p
.
1
7
5
8
1
–
1
7
5
9
2
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
5
1
0
7
0
.
2
0
2
3
.
0
1
6
1
6
.
[
4
7
]
X
.
C
h
e
n
,
Z.
Z
h
o
u
,
W
.
L
i
a
n
g
,
a
n
d
M
.
W
a
n
g
,
“
A
m
e
t
h
o
d
o
f
p
e
r
f
o
r
m
i
n
g
l
o
o
p
c
l
o
s
i
n
g
u
s
i
n
g
M
a
sk
R
-
C
N
N
m
o
d
e
l
i
n
S
LA
M
sy
s
t
e
m,”
i
n
2
0
1
9
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
M
e
c
h
a
t
r
o
n
i
c
s
a
n
d
A
u
t
o
m
a
t
i
o
n
(
I
C
MA
)
,
A
u
g
.
2
0
1
9
,
p
p
.
1
0
3
5
–
1
0
4
0
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
M
A
.
2
0
1
9
.
8
8
1
6
5
4
7
.
[
4
8
]
K
.
D
a
i
,
L.
C
h
e
n
g
,
R
.
Y
a
n
g
,
a
n
d
G
.
Y
a
n
,
“
L
o
o
p
c
l
o
s
u
r
e
d
e
t
e
c
t
i
o
n
u
s
i
n
g
K
P
C
A
a
n
d
C
N
N
f
o
r
v
i
s
u
a
l
S
LA
M
,
”
i
n
2
0
2
1
4
0
t
h
C
h
i
n
e
s
e
C
o
n
t
r
o
l
C
o
n
f
e
re
n
c
e
(
C
C
C
)
,
Ju
l
.
2
0
2
1
,
p
p
.
8
0
8
8
–
8
0
9
3
.
d
o
i
:
1
0
.
2
3
9
1
9
/
C
C
C
5
2
3
6
3
.
2
0
2
1
.
9
5
5
0
4
3
2
.
[
4
9
]
X
.
Zh
a
n
g
,
X
.
W
a
n
g
,
a
n
d
R
.
Z
h
a
n
g
,
“
D
y
n
a
m
i
c
se
ma
n
t
i
c
s
S
LA
M
b
a
se
d
o
n
i
m
p
r
o
v
e
d
M
a
s
k
R
-
C
N
N
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
Evaluation Warning : The document was created with Spire.PDF for Python.