TELKOM
NIKA
, Vol.13, No
.3, Septembe
r 2015, pp. 9
22~929
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i3.1810
922
Re
cei
v
ed Ap
ril 1, 2015; Re
vised J
une 3,
2015; Accept
ed Ju
ne 20, 2
015
Feedback Linearization Control for Path Tracking of
Articulated Dump Truck
Xuan Zhao
1
, Jue Yang*
1
, Wenming Zh
ang
1
, Jun Ze
ng
2
1
School of Mec
han
ical En
gi
ne
erin
g, Univer
s
i
ty of Scie
nce & T
e
chnolog
y B
e
ijin
g,
Beiji
ng, Ch
in
a
2
School of Co
mputer an
d Co
mmunicati
on E
ngi
neer
in
g,
Uni
v
ersit
y
of Sci
e
n
c
e &
T
e
chnol
o
g
y
B
e
ij
ing,
Beiji
ng, Ch
in
a, T
e
lp
:
+
86-010-
623
32
467
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: zhao
xu
an
120
@12
6
.com
A
b
st
r
a
ct
T
he articu
late
d du
mp truck
is a w
i
des
p
r
ead tra
n
sport
vehicl
e for
narrow
rou
g
h
terrain
envir
on
me
nt. T
o
ac
hiev
e th
e
auton
o
m
o
u
s dr
iving
i
n
the
u
n
dergr
oun
d tu
n
nel, th
is artic
l
e
pro
poses
a
p
a
th
follow
i
n
g
strategy -for articu
lated v
ehic
l
e
base
d
on
fe
e
dback l
i
n
eari
z
ation
alg
o
rith
m. F
i
rst of all
,
the
kine
matic mo
d
e
l of articulat
e
d vehicl
e, w
h
ich refl
ects the relati
onsh
i
p b
e
tw
een the structure para
m
eter
s
and state vari
a
b
les, has b
e
e
n
establis
hed. R
e
ferrin
g
to
the mo
de
l, the non
line
a
r errors e
quati
on b
e
tw
een
real
path
an
d r
e
ferenc
e p
a
th,
w
h
ich are
as t
he fe
edb
a
ck fr
om the
path tr
ackin
g
pr
ocess
,
has b
e
e
n
so
l
v
ed
and
li
near
i
z
e
d
.
After esti
mati
ng th
e syste
m
control
l
a
b
il
ity, the
path fo
llo
w
i
ng co
ntroll
er
w
i
th feed
bac
k
line
a
ri
z
a
t
i
on al
gorith
m
h
a
s
b
een desi
g
n
ed throug
h
calc
ul
ating th
e par
a
m
eters w
i
th th
e pol
e assi
gn
me
nt
accord
ing to t
he err
o
r eq
ua
tion. F
i
na
lly, the H
a
rdw
a
re-I
n-the-L
o
o
p
si
mu
lati
on o
n
NI cRIO and
PXI
control
l
er
has
bee
n -co
n
d
u
cted for v
e
rifyin
g
the co
nt
rol
qu
ality a
nd r
e
a
l
-time
path
tracki
ng p
e
rfor
ma
nc
e.
T
he res
u
lt sh
o
w
s that the
pat
h tracki
ng c
ont
roller
w
i
th fee
d
back
lin
eari
z
at
i
on c
an track
th
e refer
ence
p
a
t
h
accurately.
Ke
y
w
ords
: arti
culate
d veh
i
cle
,
path tracking,
feedback l
i
n
e
a
r
i
z
at
io
n, hardw
are-In-the-
loo
p
Copy
right
©
2015 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
The a
r
ticulate
d dum
p truck is
a tra
n
spo
r
t vehicle
which is wi
dely u
s
ed
in mi
ning
, water
con
s
e
r
van
c
y and co
nst
r
u
c
tion. It esp
e
cially ad
apt
s to the narrow spa
c
e, rough terrai
n and
severe we
ath
e
r, for exam
ple, the tunn
el in
the und
erg
r
ou
nd min
i
ng. The auto
nomou
s d
r
iving
system, an
advan
ced a
u
t
omatic tech
nology, ca
n
improve the
vehicle p
r
o
d
u
ctivity, operator
safety an
d ex
hau
st cl
eanli
n
ess [1]. Th
e p
u
rpo
s
e
of thi
s
arti
cle i
s
to
find a
path
foll
owin
g
strateg
y
for autono
mo
us drivin
g system on articul
a
ted vehicl
e.
Before
devel
oping
the
p
a
th follo
wing
stra
tegy, th
e a
r
ticulate
d
vehicl
e m
o
del, path
followin
g
algo
rithm and
con
t
roller te
st me
thod nee
d to be cle
a
r. The
articul
a
ted ve
hicle mo
delin
g
and
path follo
wing
control
appli
c
ation
h
a
ve be
en inv
e
st
igate
d
in
many stu
d
ie
s. In [2-4]
kine
matic
model of
arti
culate
d vehi
cle and
error
model b
e
twe
en real an
d referen
c
e
pat
h are presen
ted,
and the
pat
h tra
cki
ng
si
mulation
with
model
pre
d
i
c
tive co
ntrol
is a
pplied,
while i
n
[5]
the
feedba
ck co
n
t
roller b
a
sed
on Lyapu
nov
appro
a
ch is
desi
gne
d an
d the stability of the close
d
-
loop sy
stem i
s
proved in t
heory. The
si
mulation
s
in these literatu
r
es a
r
e no
n-re
al-time an
d the
real
-time pe
rforme
nce ha
s not been ve
rified. More
over
, in [6] a trajectory tra
c
king
stratea
g
y fora
new
stru
ctu
r
e automate
d
guided ve
hi
cle i
s
presen
ted, while in
[7] a feedb
ack linea
ri
zat
i
on
control fo
r al
most
global
o
u
tput
-feed
ba
ck trackin
g
i
s
p
r
ovided
for th
e un
deractu
ated a
u
tonom
o
u
s
quad
roto
r. Bo
th the literatu
r
es have
de
si
gned th
e
con
t
roller
of feed
back lin
eari
z
ation. Ho
wev
e
r,
the model
s o
f
the plants a
r
e di
stinct fro
m
the arti
cul
a
ted vehicl
e. This a
r
ticle
n
eed to de
sig
n
a
path follo
win
g
algo
rithm
controlle
r
with
a
kinem
atic
model
of arti
culate
d vehi
cle and
test i
n
a
real
-time envi
r
onm
ent.
An articulate
d dump
tru
c
k con
s
i
s
ts of
a tra
c
tor, a trailer a
nd a
n
articul
a
ted b
o
d
y. This
stru
cture h
a
ve two
de
gre
e
s
of f
r
eed
om,
yaw
and
roll,
for a
sho
r
ter stee
ring
radi
us
and
keepi
ng
all tire
s
cont
acting
the
ground
on
the
roug
h te
rrai
n
respe
c
tively.
Ho
wever, co
mpared with the
traditional A
c
kerman
n ste
e
ring m
e
cha
n
ism, this
a
r
ticulated ste
e
ring stru
ctu
r
e
has com
p
l
e
x
steeri
ng
process. To
solve
this
problem,
the m
a
thema
t
ical mo
del
of
arti
culated
vehicl
e i
s
d
e
riv
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Feedb
ack Lin
eari
z
ation
Co
ntrol for Path
Tra
cki
ng of Articulate
d Du
m
p
Truck (Xu
an Zhao
)
923
in whi
c
h
the v
e
locity of the
front fram
e i
s
con
s
id
ere
d
a
s
a
mea
s
u
r
ab
le varia
b
le. F
u
rthe
rmo
r
e, t
he
articul
a
ted ve
hicle mo
del,
as a hig
h
ly nonline
a
r m
o
d
e
l, is hard to control.
Ho
we
ver, the feedb
ack
lineari
z
atio
n can simplify
the
controll
er desi
gn.
Mea
n
w
hile, control
l
er
real-tim
e perfo
rmen
ce is
importa
nt, while tradition
al offline si
mulation me
thod ca
nnot
test. Hard
ware-In
-
the
-
L
oop
simulatio
n
is thus
con
d
u
c
te
d toverify the real
-time pe
rforma
nce [8].
In this
articl
e, a 3
5
ton
s
un
derg
r
o
und
art
i
cu
lated mining truc
k with
elec
tical transmis
s
i
on
desi
gne
d by
University of
Scien
c
e &
Te
chn
o
logy Be
ij
ing i
s
sele
cte
d
a
s
the
re
se
arch p
r
ototyp
e.
In Section 2,
a kine
matic
model of the
articu
l
a
ted ve
hicle h
a
s
bee
n builtthro
ugh
the geomet
ri
cal
relation
shi
p
. In addition, th
e definition o
f
the errors b
e
twee
n re
al
path an
d refe
ren
c
e p
a
th h
a
s
been
discri
be
d and
modell
ed. In Sectio
n 3,acco
rd
in
g
to the feedb
ack line
a
ri
zat
i
on metho
d
, the
state equatio
n and output equatio
n of the nonline
a
r ar
ticulated vehi
cle have bee
n transfo
rme
d
to
a linear
controllable an
d o
b
se
rvable
system. Then
t
he co
ntrolle
r has b
een co
nfigure
d
by linear
control metho
d
to control the model for t
he pre
c
i
s
e pa
th tracking. In
Section 4, the Ha
rdware-I
n-
the-Lo
op sim
u
lation devices have be
e
n
sho
w
na
nd
the simulati
on pro
c
e
dure with different
platform
s h
a
ve be
en int
r
od
uce
d
. In Se
ction 5, th
e si
mulation
ha
s bee
n lau
n
ch
ed to ve
rify the
quality of the
controlle
r. Th
e re
sult
s h
a
ve be
en
sho
w
n with
the g
r
a
phs. An
d the
interp
retation
of
the results ha
s bee
n discu
s
s. In Section
6, the
con
c
lu
sion p
r
e
s
ent
s the findings
of this article.
2. Articula
te
d Vehicle Kinemics Mod
e
ling
2.1. Articula
ted Vehicle Mathema
t
ical
Model
The arti
culat
ed vehicl
e turning
state is
pre
s
ented i
n
Figure 1. In
this figure,
O
is the
instanta
neo
u
s
cente
r
of m
o
vement.
P
f
(
x
f
,
y
f
) an
d
P
r
(
x
r
,
y
r
), de
note th
e co
rres
pon
di
ng center poi
nts
of trac
tor and trailer.
l
f
an
d
l
r
are the l
e
n
g
th of the fro
n
t and rea
r
u
n
its.
θ
f
an
d
θ
r
denote th
e u
n
its
orientatio
n.
γ
is
th
e a
r
tic
u
late
d
a
n
g
l
e
w
h
ic
h
is
d
e
f
in
e
d
a
s
th
e
d
i
ffe
re
n
c
e
be
tw
e
e
n
th
e
fro
n
t
and
rea
r
ori
entati
on. Usually, con
s
id
erin
g simplifying cal
c
ulatio
n,
P
f
is the whole v
ehicl
e refe
re
nce
point, be
cau
s
e the velo
ci
ty
v
f
orientati
on of this p
o
int is coin
ci
dent with th
e wh
ole vehi
cle
orientatio
n [9]. The velocity of articulated
vehicle is d
e
fined a
s
f
vv
(1)
Figure 1. Articulate
d vehicl
e scheme
The co
ordinat
es of
P
f
is
:
co
s
sin
f
ff
f
ff
xv
yv
(2)
The rate
of
θ
f
is compo
s
e
d
of th
e a
n
g
u
lar velo
city in con
s
tant
ra
dius turning
and ya
w
veloc
i
ty in fix
ed
P
f
pivot steerin
g.
si
n
cos
f
r
f
f
r
vl
ll
(3)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
9
30
TELKOM
NIKA
Vol. 13, No. 3, September 20
15 : 922 – 929
924
The arti
culate
d vehicle
state equatio
n
P
f
=[
x
f
,
y
f
,
θ
f
,
γ
] is
:
0
cos
0
sin
si
n
co
s
c
o
s
01
f
f
f
f
r
f
fr
fr
x
y
l
v
ll
ll
(4)
2.2. Path De
scription
Figure 2
sh
o
w
s th
e e
r
rors bet
wee
n
th
e re
al an
d referen
c
e
pat
h
. The
small
circle
of
whi
c
h th
e
ce
nter i
s
c
i
s
th
e re
al vehi
cle
path, while th
e big
ci
rcl
e
o
f
whi
c
h th
e
center is
C
is
the
referen
c
e pat
h. Ideally, the vehicle shoul
d pass
P
1
,
P
2
,
P
3
. The variables a
r
e defin
ed as:
a) Lateral
di
spla
cem
ent error
d
:the
l
a
teral displ
a
cement errorb
etwee
n
the
vehicle
referen
c
e poi
nt
p
and the correspon
ding
point
P
(the n
eare
s
t point o
n
the referen
c
e path
)
;
b) Ori
entation
erro
r
:the o
r
ientation e
r
ror between t
h
e velocity orientation of
p
and the
tangential o
r
i
entation of
P
;
c) Curvatu
r
e error
c
: the curvature error bet
wee
n
the curvatu
r
e of
p
and
P
.
Figure 2. Articulate
d vehicl
e plan-vie
w
2.3. Error Modeling
Figure 3 is
a
part of Figu
re 2. When t
he vehicl
e m
o
ves fro
m
p
to
p’
, i
t
revolves
d
θ
arou
nd in
stan
taneou
s
cent
er
c
at radiu
s
r
. Ra
dial lin
e
s
cp
a
nd
Cp’
intersec
t the
referenc
e path
cir
c
le at
P
an
d
P’
respectiv
e
ly. The angl
e betwe
en
CP
and
CP’
is denote
d
d
.
Figure 3. Geo
m
etric
relatio
n
shi
p
of errors
a) Late
r
al di
splacement e
r
ror
d
A
ssu
ming b
o
t
h
d
θ
a
nd d
a
r
e sm
all angl
es, the ch
ang
e of lateral di
spla
cem
ent e
rro
r is
d
dr
d
(5)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Feedb
ack Lin
eari
z
ation
Co
ntrol for Path
Tra
cki
ng of Articulate
d Du
m
p
Truck (Xu
an Zhao
)
925
Subs
tituted with
v
=
r
d
θ
/d
t
, then the rate
of lateral disp
lacem
ent erro
r is:
d
v
(6)
b) Ori
entation
erro
r
Acco
rdi
ng tothe geom
etri
c relation
shi
p
,
dd
d
(7)
()
d
Rd
r
d
(8)
From e
quatio
n(7
)(8
), the st
eady-state st
eer
in
gdifferen
t
ial of orientation error i
s
1
d
r
dd
R
(9)
Gene
rally, th
e wi
dth of th
e tunn
el is a
bout
5
m a
n
d
wi
dth of th
e vehi
cle i
s
3.4 m. So
d
≤
0.6 m, whi
l
e the v
ehicle steeri
ng
radiu
s
r
≤
6.6 mthat
mean
s
R
≥
6.6
m. Thus, assume that
R
≫
d
.
Andco
n
si
de
ring the a
dditio
nal yaw
angl
e velocity wh
en
≠
0, as
wel
l
as
sub
s
tituting for
v
=
r
d
θ
/d
t
and
c
=
r
-1
-
R
-1
, the rate of orientation erro
r is:
co
s
r
c
rf
l
v
ll
(10
)
c) Curvatu
r
e error
c
Known vehi
cl
e velocity
v
a
nd refe
ren
c
e
path radi
us
R
, the real path
radiu
s
is:
co
s
sin
rf
fr
ll
v
rv
vl
(11
)
Differentiatin
g
re
cip
r
o
c
al of
Equation (1
1
)
with re
sp
ect
to time
t
give
s,
1
2
22
(c
o
s
)
(
c
o
s
)
(
s
i
n
)
(c
o
s
)
(
c
o
s
)
fr
r
r
f
f
r
c
rf
rf
dr
vl
l
l
l
l
l
l
dt
v
l
l
v
l
l
(12
)
From Equ
a
tio
n
(6), (10), (1
2), the linea
ri
zed
state equ
ation is:
1
1
1
0
00
0
00
c
o
s
0
000
co
s
co
s
dd
rr
f
cc
rr
f
rf
v
vl
l
l
lv
l
l
ll
(13
)
Known
-
0.2
5
π
<
γ
<0.
2
5
π
fro
m
vehicle
structure, an a
s
sumptio
n
is
made that
γ
is a smal
l
angle
mea
s
u
r
ed
in
radi
an
s a
nd
L
=
l
f
+
l
r
.
Re
define
Eq
uation
(13
)
a
s
the
arti
culat
ed vehi
cle
st
ate
equatio
n for p
a
th tracking S
M
C de
sign.
1
1
00
0
00
00
0
dd
r
cc
v
vl
L
L
(14
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 3, September 20
15 : 922 – 929
926
3. Feedba
ck
Linearizatio
n Control Al
gorithm De
sign
3.1. Feedba
c
k
Linearizati
on Metho
d
Feedb
ack li
n
eari
z
ation
is
a commo
n m
e
thod
us
e
d
in
co
ntrollin
g n
online
a
r
syst
ems. T
he
approa
ch inv
o
lves
comin
g
up with
a tra
n
sformati
on
of the nonli
n
ear
system i
n
to an e
quiv
a
lent
linear
system
through a
ch
ange of varia
b
les an
d a
su
itable cont
rol
input. The st
ate feedba
ck is
implimente
d
on the ba
sis of vehicle ki
nematic
m
o
d
e
l.The nonli
n
ear
kinem
atic model can
be
transfo
rme
d
i
n
to a cl
osed-loop line
a
r
sy
stem vi
a introdu
cing a
p
p
r
opriate
state
feedba
cks. T
he
well-esta
blish
ed line
a
r
con
t
rol theo
rie
s
are the
n
ap
pl
icabl
e for the
nonlin
ear
sy
stem
s with
st
ate
feedba
ck.
More
over, th
e feedb
ack li
neari
z
e
d
sy
stems a
r
e ob
servable as well
as co
ntrol
l
able.
A
linear
cont
rol
method is the
n
prop
osed in
or
de
r to cont
rol the vehicl
e kinem
atic
model.
3.2. Contr
o
ller Desig
n
Before designing the
cont
roller, the system
cont
rollability must be estimated. If the
r
th
derivative
of system can e
x
pres
s the
rel
a
tionship bet
wee
n
output
y
(
t
) and
in
put
u
(
t
),
r
i
s
defi
ned
as the relative degree. Th
e
system i
s
controlla
ble when
r
≤
n
where
n
is syste
m
degree [11].
Analytically, from eq
uation
(
14
),
c
can b
e
controlled b
y
and
can
be co
ntrolle
d by
and
c
as
w
e
ll a
s
d
can be
cont
rolled by
. Thus, the
articul
a
ted v
ehicl
e sy
stem
is controll
abl
e
whe
n
the a
r
t
i
culate
d an
gl
e rate
is a
s
the input. A
s
suming
that
the state v
a
riabl
e vecto
r
x
=[
d
,
,
c
]
T
i
s
mea
s
u
r
able,
the path
tra
c
king
controll
e
r
can
be
de
signed
with th
e
state va
riabl
e
feedba
ck [12].
Theo
retically,
the clo
s
ed
-lo
op system ca
n
be
formatte
d a
s
wish.
T
he
state fee
d
back i
s
denote
d
a
s
u
=-
Kx
, where
K
is feed
ba
ck gain
and
K
=[
k
1
,
k
2
,
k
3
]. The
optimization
obje
c
tive is
to
find an app
ro
priate
K
that can le
ad any
arbitrary
x
to the des
i
red value in time [13].
Suppo
sed th
at the object
controll
ed is a linear
time-inva
r
iant time system e
x
presse
d
with state spa
c
e eq
uation
s
as follo
w[14]
x
Ax
B
u
(15
)
The linea
r fee
dba
ck i
s
:
Kx
u
(16
)
()
x
AB
K
x
(17
)
Whe
r
e
0
0
0
0
0
0
0
v
v
A
,
1
1
0
r
Bl
L
L
,
c
d
x
,
u
,
12
3
[,
,
]
Kk
k
k
.
The
sy
stem transitio
n
pe
rforman
c
e sh
ould
be co
n
s
ide
r
ed, su
ch
as re
spo
n
s
e
time,
setting time
and oversh
o
o
t. The pole
assi
gnme
n
tis the sol
u
tio
n
. In this article, different
ial
transfo
rmatio
n method is
use
d
for assi
gning the pol
e to the arbitrary location
on plane
S
for
system
stabili
ty and tran
sition pe
rform
a
n
c
e. So
the cl
ose
d
-lo
op p
o
le can
be
set on the de
sire
d
locatio
n
in
order to
ke
ep t
he sy
stem h
a
v
ing 2nd
ord
e
r dyn
a
mic resp
on
se that
natural
freq
u
ency
ω
n
and damp
i
ng ratio
ξ
i
s
d
o
minant. Me
anwhile, the 3rd pol
e sho
u
ld be far a
w
ay for these t
w
o
pole
s
on the l
e
ft side half o
pen complex
plane
S
[15].
Considering t
he
closed-l
oop stability of this
system, t
he eigenvalue
s
of the ei
g
enmatrix
A
-
BK
is on the
left side of plane
S
[16].
32
2
23
12
1
()
(
)
0
r
r
lk
k
lk
k
k
ss
v
s
v
LL
L
(18
)
The t
r
u
ck si
z
e
s ar
e
l
f
=1.6
8m and
L
=5.1
2m and it drives with the con
s
tant velo
city
v
=3
m/s. Solving the Equation
(18), the
State feedba
ck g
a
i
n
can b
e
obta
i
n as:
12
3
[
,
,
]
[
0
.7
,
3
.9
,
1
5
.
6
]
Kk
k
k
(19
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Feedb
ack Lin
eari
z
ation
Co
ntrol for Path
Tra
cki
ng of Articulate
d Du
m
p
Truck (Xu
an Zhao
)
927
4. Hard
w
a
re
-In-the
-
Loo
p Simulation
Hardware-In
-
the-L
oop
is a
form
of re
al
-tim
e
simulati
on. Hard
wa
re-In-th
e-L
oop
differs
from pu
re
rea
l
-time si
mulat
i
on by the
ad
dition of
a
“re
a
l”
comp
one
n
t
in the loo
p
.This
com
pon
ent
may be an el
ectro
n
ic
co
ntrol unit (ECU for autom
ot
ive, FADEC for
Aero
spa
c
e
)
o
r
a re
al engi
n
e
.
The p
u
rp
ose
of a
Hardware
-In-the
-
L
o
op si
mulatio
n
is to
provid
e all of the
electri
c
al
sti
m
uli
need
ed to fu
lly exerci
se t
he ECU. In
effect, “foo
lin
g” the
ECU i
n
to thinki
ng t
hat it is i
nde
ed
c
o
nnec
ted to
a real
plant.In this
artic
l
e t
he comp
onent is
NI cRIO as
a
path following
c
ont
roller.
In this
articl
e
,the Ha
rd
wa
re-In-th
e-L
oop
devices
are
sh
own in
Fi
gure
4
(
a) an
d flowch
art i
s
in
Figure 4(b
)
. It shows that the plant, articulate
d
vehicle modelled
by MapleSim
, is simulated
in
PXI and the cRIO
cont
roll
er, in whi
c
h
prog
ram i
s
compiled by L
abView, is
re
al. To observ
e
dire
ctly, all the data i
s
u
p
lo
aded to
a P
C
to di
splay wi
th a g
r
aphi
cal
use
r
inte
rfa
c
e programme
d
by LabView.
(a) Simul
a
tio
n
dev
ice
s
Pa
t
h
Follow
i
ng
Co
n
t
ro
l
L
a
w
Vi
a
La
b
v
ie
w
Ar
t
i
c
u
l
a
t
e
d
V
e
hi
cl
e M
o
del
Vi
a
Map
l
e
S
i
m
Da
t
a
M
o
n
i
to
r
Via
La
b
v
i
e
w
Co
n
t
r
o
l
Seq
u
e
n
c
e
St
a
t
u
s
F
eed
b
a
c
k
Co
n
t
ro
l
Seq
u
en
c
e
St
a
t
u
s
F
eed
b
a
c
k
(b) Simulation flowchart
Figure 4. Hardwa
re
-In-the
-
Loop
simul
a
tion
5. Results a
nd Discu
ssi
on
Duri
ng the si
mulation, the truck follows
a circle path
with radi
us
r
=25 m in the consta
nt
spe
ed
v
=3 m
/
s. The
ori
g
in
al poi
nt of th
e glo
b
le
coo
r
dinate syste
m
is on
th
e centre of
the circle
path. The sta
r
ting point i
s
(-3,-2
5)
, an
d the initial dire
ction is -
∞
of X
axle.The sim
u
lation du
rati
on
is 100
se
cen
d
s.
As Figu
re
5 a
nd Fig
u
re
6
shown, the rea
l
pat
h i
s
almo
st coi
n
ci
dent
with refere
nce path.
Con
s
id
erin
g t
heste
erin
g
wheel a
ngle
is
the input
va
ra
ible on
the m
anne
d vehi
cl
e, and i
n
o
r
d
e
r to
sho
w
th
e veh
i
cle
stee
ring
pro
c
ed
ure
cl
early, the
de
sire
d
articulat
ed a
ngle
γ
,int
egratio
n of
, is
as the inp
u
t of the articul
a
ted vehicl
e
model in
stea
d of articul
a
ted angl
e rate
. All
the gra
p
h
s
has oversh
oo
t in Figu
re
6
before
10
secend
s, be
ca
use the ve
hicl
e
need
s to
a
c
celerate
from
0 to
3 m/s
and t
he initial
errors a
r
e
relat
i
ve larg
e.
Desired
articul
a
ted an
gle f
r
om
co
ntroll
er
approa
che
s
t
o
0.21
rad
(1
2°) in
10
se
cend
s. Th
e d
e
s
ire
d
a
r
ticulat
ed a
ngle
is calcul
ated
by
d
,
and
c
who
s
e chan
ging
trend
s are n
o
t
identical.
Whe
n
γ
le
ad
s th
ree
varia
b
les chan
ge
s to
different dire
ction
s
, the ch
attering ap
pe
ars
while it
is slight.Lateral
displa
cem
e
n
t
error
red
u
ces
respe
c
tively from 200 mm
to 100mm after the sh
ort o
v
ersh
oot. Ref
e
rri
ng to the front wh
eel tra
ck
2280 mm, the
erro
r is only 4%. Orientation error st
ay
s near 0.01 ra
d(0.5°
), and curvature error is
almost 0. All
the variabl
e
s
finally tend
to be sta
b
le
in 10 secen
d
s
with a little ch
attering
for
adju
s
ing. T
h
e
re
sult
sh
ows
that
the fe
edb
ack lin
eari
z
ati
on
cont
rolle
r
can
tra
c
k the
referen
c
e
pat
h
effectively in real-time e
n
vironment.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 3, September 20
15 : 922 – 929
928
Figure 5. Con
t
rast bet
ween
refere
nce an
d real path
(a)
De
sire
d articulated a
ngl
e
γ
(b) L
a
teral di
s
p
lac
e
ment e
r
r
o
r
d
(c) Ori
entatio
n error
(d) Curv
atu
r
e
erro
r
c
Figure 6. Articulate
d angl
e
and errors chang
e
5. Conclusio
n
In this arti
cle,
the re
sea
r
ch
prototype i
s
a 35-to
nne el
ectri
c
al transmissi
on u
nde
rgroun
d
mining a
r
ticul
a
ted dump truck. And the hard
w
a
r
e in
t
he loop
simul
a
tion is ba
se
d on NI cRIO
and
PXI controlle
r. The feedba
ck linea
rization
is us
e
d
for d
e
sig
n
ing the
path tra
ckin
g
controlle
r.
The co
ncl
u
si
ons of this a
r
t
i
cle are:
a)
For a
r
ticulat
ed vehi
cle
model, a
highly nonli
n
e
a
r mo
del,th
e feedb
ack lineari
z
atio
n
controlle
rcant
rack the
referen
c
e
pa
th accu
ratel
y
. Both the dynami
c
and
ste
a
d
y
cha
r
a
c
teri
sticscan fulfill the demand.
b) Th
e feedb
ack linea
ri
zat
i
on co
ntroll
er devel
ope
d
by the kine
m
a
tics m
odel
can
cont
rol t
h
e
vehicle
to foll
ow th
e
refe
re
nce
path
with
out a
c
cu
rate
dynamic mo
d
e
l. The
real
p
a
th is
smooth
and little chat
tering.
c)
Com
p
a
r
ed
with the
rea
l
vehicle te
st
, Hardware
-I
n-the
-
Lo
op
si
mulation i
s
e
c
on
omical an
d
efficient. And the real-ti
m
e perfo
rm
ance is better than the
off-line sim
u
lation. The
cha
r
a
c
teri
stic of the control
l
er is te
sted comprehe
nsiv
ely.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Feedb
ack Lin
eari
z
ation
Co
ntrol for Path
Tra
cki
ng of Articulate
d Du
m
p
Truck (Xu
an Zhao
)
929
Ackn
o
w
l
e
dg
ements
This work was finan
cially
supp
orted b
y
the Nation
al High Te
ch
nology Rese
arch and
Develo
pment
Program
(8
63 Prog
ram
)
of China, unde
r Awa
r
d
s
2011AA0
6
0404 Intellig
ent
Und
e
rg
ro
und Mining
Truck.
Referen
ces
[1]
Dragt BJ, Camisani-C
alzolari FR, Craig IK.
An Overvie
w
of the Auto
mati
on
of Lo
a
d
-Ha
u
l-D
u
mp
Vehic
l
es
in
an
Un
dergr
ou
nd
Mini
ng
Env
i
ron
m
e
n
t
.
Proce
edi
ngs
of th
e
17th
W
o
rl
d
Con
g
ress th
e
Internatio
na
l F
eder
ation
of Automatic Contro
l
. Prague. 20
05
; 16: 1388-
140
0.
[2]
Na
yl T
,
Nikolak
opo
ulos G, Gu
stfsson T
.
Sw
itchin
g Mod
e
l Pr
edictiv
e Co
ntro
l for an Artic
u
l
a
ted Veh
i
cl
e
und
er V
a
ryin
g
Sli
p
A
ngl
e
.
20th M
e
d
i
terr
ane
an
Co
nfer
ence
o
n
C
o
n
t
rol & A
u
tom
a
tion
(MED).
Barcel
ona. 2
0
1
2
: 890-8
95.
[3]
Lee J
H
, Yo
o
W
S
. Predictive
Contro
l of
a
Vehic
l
e
T
r
aject
o
r
y
Usi
ng
a C
oup
led
Vector
w
i
t
h
Ve
hicl
e
Veloc
i
t
y
a
nd Si
desli
p Ang
l
e.
Internati
o
n
a
l Jo
urna
l of Automotive T
e
chn
o
l
o
gy
. 2009; 1
0
(2)
:
211-21
7.
[4]
Ridl
e
y P, C
o
rk
e P. L
oad
Ha
u
l
Dum
p
Ve
hic
l
e Kin
e
matics
a
nd C
ontro
l.
Jo
urna
l of Dy
na
mic
Syste
m
s,
Measur
e
m
ent, and C
ontrol
. 2
003; 12
5(1): 54
-59.
[5]
Petrov P, Chaky
r
ski D.
Path F
o
llow
i
ng
Contro
l of an
Articulated M
i
nin
g
Veh
i
cl
e via Lya
pun
o
v
T
e
chni
ques
. 17
th
NNT
K Internatio
nal C
onfer
enc
e. Sofia. 2
0
09: 396-
40
0.
[6]
Ye
X, W
u
Z
,
Z
hao F
.
R
e
se
arch
on
a 3
D
O
F
Automated
Guid
ed V
e
h
i
c
l
e b
a
se
d o
n
t
he I
m
pr
oved
F
eedb
ack Li
ne
ari
z
a
t
i
on Meth
od
. 33rd C
h
i
n
e
s
e Contro
l Con
f
erence (C
CC).
Nanji
ng. 2
014:
184-1
88.
[7]
Maithrip
al
a DH
S, Berg JM.
R
obust T
r
ackin
g
Control for Under
act
uate
d
Autono
mous V
ehicl
es Usin
g
F
eedb
ack Lin
e
a
ri
z
a
ti
on
. Proc
eed
ings
of
IEE
E
/ASME Intern
ation
a
l
Co
nfer
ence
on
Adv
a
n
c
ed Inte
lli
ge
nt
Mechatro
nics (
A
IM). Besanco
n
. 2014: 4
46-4
51.
[8]
Jackson
R, Sarw
ar S. PX
I
Ad
dresses Ne
w
HIL
App
licati
o
n
s
.
Evaluati
on E
ngi
neer
in
g
. 20
05; 44(
6): 24
-
28.
[9]
Pei
XZ
, Li
u Z
Y
, Pei R. A
ppl
ic
ation
of E
x
act
F
eedb
ack L
i
ne
arizati
on to T
r
ajector
y
T
r
acki
ng for M
obi
le
Robot.
Robot
. 200
1; 23(7): 66
1-68
0.
[10]
Yuan L. Missil
e
Attitude Co
nt
rol S
y
stem
Based o
n
T
r
ajector
y
L
i
n
eariz
ation. PhD T
hesis. Harbi
n
:
H
a
rb
in
En
gi
nee
ri
ng
U
n
i
v
e
r
si
ty
; 2
0
1
3
.
[11]
Mahi
ndrak
ar A
D
, Ban
a
var R
N
.
Contro
ll
abi
li
ty Properti
es o
f
a Pla
nar
3R
Und
e
ractuat
ed
Mani
pu
lator
.
IEEE Confere
n
c
e on Co
ntrol
Appl
ic
atio
ns. Glasg
o
w
.
20
02: 489-
494.
[12]
Akhtar A, Niel
s
en C.
Path F
o
llow
i
ng for a
Car-lik
e Ro
bot
Using T
r
a
n
sv
erse F
e
e
dback
Line
ari
z
at
io
n
and T
a
nge
ntia
l
Dyna
mic Exte
nsio
n
. Procee
d
i
ngs of the IE
EE Conf
er
enc
e on D
e
cisi
on
and C
ontro
l.
Orland
o. 201
1: 7974-
79
79.
[13]
Song
XJ.
Des
i
gn and
Simu
la
tion of
PM
SM
F
eed
back
Lin
eariz
ation
Co
n
t
rol S
y
stem.
Te
lkomn
i
ka -
Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2013; 1
1
(3): 1
245-
125
0.
[14]
Liu
D, Z
hou
D, Liu Y.
A M
e
th
od of Mu
lti-rate
Sing
le-si
de
N
e
tw
orked co
ntrol for L
i
n
ear T
i
me-Inv
ari
ant
System w
i
th State-feed
back
. 5th Internati
ona
l Conf
eren
ce on
Com
p
u
t
er Science
a
nd Educ
atio
n
(ICCSE). Hefei
.
2010: 12
70-1
274.
[15]
Hardi
ans
ya
h,
Juna
idi. Multi
obj
ective H2/
H
∞
Contro
l Desig
n
w
i
th
Regi
on
al Po
le
Constrai
nts.
T
E
LKOMNIKA T
e
leco
mmunic
a
tion C
o
mputi
n
g Electron
ics a
nd Co
ntrol
. 20
12; 10(1): 1
03-
112.
[16]
Man Z
,
Pala
ni
s
w
a
m
i M. Ro
b
u
st T
r
a
cking C
ontrol for
Rig
id
Rob
o
tic Man
i
pul
ators.
IEEE Transactions
on Auto
matic
Contro
l
. 199
4; 39(1): 15
4-1
5
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.