This art icle was downloaded by: [ I st anbul Technical Universit y]

On: 03 Oct ober 2012, At : 06: 14

Publisher: Taylor & Francis
I nforma Lt d Regist ered in England and Wales Regist ered Number: 1072954 Regist ered office: Mort imer House,
37- 41 Mort imer St reet , London W1T 3JH, UK
Int ernat i onal Journal of Cont rol
Publ i cat i on det ai l s, i ncl udi ng i nst r uct i ons f or aut hor s and subscr i pt i on i nf or mat i on:
ht t p: / / www. t andf onl i ne. com/ l oi / t con20
Tracki ng cont rol of t ri m t raj ect ori es of a bl i mp f or
ascent and descent f l i ght manoeuvres
L. Bej i
a
& A. Abi chou
b
a
Labor at oi r des Syst emes Compl exes, LSC, CNRS- FRE2494, Uni ver si t d Evr y Val
d Essonne, FRANCE 40, r ue du Pel voux, 91020, Evr y, Cedex
b
Labor at oi r e d i ngni er i e mat hmat i que, LI M, Ecol e Pol yt echni que de Tuni si e, BP 743,
2078 La Mar sa, Tuni si a
c
Labor at oi r des Syst emes Compl exes, LSC, CNRS- FRE2494, Uni ver si t d Evr y Val
d Essonne, FRANCE 40, r ue du Pel voux, 91020, Evr y, Cedex E- mai l :
Ver si on of r ecor d f i r st publ i shed: 19 Aug 2006.
To ci t e t hi s art i cl e: L. Bej i & A. Abi chou ( 2005) : Tr acki ng cont r ol of t r i m t r aj ect or i es of a bl i mp f or ascent and descent f l i ght
manoeuvr es, Int er nat i onal Jour nal of Cont r ol , 78: 10, 706- 719
To l i nk t o t hi s art i cl e: ht t p: / / dx. doi . or g/ 10. 1080/ 002071705000118643
PLEASE SCROLL DOWN FOR ARTI CLE
Full t erms and condit ions of use: ht t p: / / www. t andfonline. com/ page/ t erms- and- condit ions
This art icle may be used for research, t eaching, and privat e st udy purposes. Any subst ant ial or syst emat ic
reproduct ion, redist ribut ion, reselling, loan, sub- licensing, syst emat ic supply, or dist ribut ion in any form t o
anyone is expressly forbidden.
The publisher does not give any warrant y express or implied or make any represent at ion t hat t he cont ent s
will be complet e or accurat e or up t o dat e. The accuracy of any inst ruct ions, formulae, and drug doses should
be independent ly verified wit h primary sources. The publisher shall not be liable for any loss, act ions, claims,
proceedings, demand, or cost s or damages what soever or howsoever caused arising direct ly or indirect ly in
connect ion wit h or arising out of t he use of t his mat erial.
International Journal of Control
Vol. 78, No. 10, 10 July 2005, 706719
Tracking control of trim trajectories of a blimp for
ascent and descent ight manoeuvres
L. BEJI*{ and A. ABICHOU{
{Laboratoire des Systemes Complexes, LSC, CNRS-FRE2494, Universite dEvry Val dEssonne,
FRANCE 40, rue du Pelvoux, 91020, Evry, Cedex
{Laboratoire dinge nierie mathe matique, LIM, Ecole Polytechnique de Tunisie, BP 743, 2078 La Marsa, Tunisia
(Received 15 May 2004; in nal form 31 March 2005)
A blimp is a small airship that has no metal framework and collapses when deated. It belongs
to family of unmanned aerial vehicles (UAVs). In this paper we address the problem of design-
ing tracking feedback control of an underactuated autonomous UAV. The ascent and descent
ight conditions as one in which the rate of change (of magnitude) of the airships state vector
is zero and the resultant of the applied forces and moments is constant lead to trimmed ight
trajectories. The subject of the tracking control is to stabilize the engine around the planned
ight. Using a combined integrator backstepping approach and Lyapunov theory, the stability
results are local and overcome the minimum number of actuators (inputs) with respect to the
blimps six degrees of freedom. Considering physic limits in UAVs, other trimmed ights are
investigated and compared.
1. Introduction
The feedback control of automatic ying machines has
attracted the attention of many researches in the past
few years. Due to inherent tasks of these engines, they
must cooperate with a minimum of load onboard and
with the maximum of autonomy. The load of the
system can be reduced by decreasing the number of
actuators. In counter balance, this will complicate the
autonomy (control) in ight due to the presence of six
degrees of freedom in UAVs. Generally, autonomous
systems are characterized by the fact that there
are more degrees of freedom than actuators, known
as under-actuated systems. Under-actuated systems
regroup indoor/outdoor blimps (airships) (De Paiva
et al. 1999, Khoury and Gillet 1999, Hygounenc 2000,
Bestaoui and Hamel 2000, Beji et al. 2003), the four-
rotor vertical take o and landing (VTOL) named as
X4-yer (Pound et al. 2002, Hamel et al. 2002), the
quad-rotor helicopter (Altug et al. 2002), the planar
vertical take o and landing (PVTOL) aircraft (Hauser
et al. 1992, Fantoni and lozano 2001, Marconi et al.
2002), and unmanned underwater vehicles (UUVs)
(Fossen 1994, Boskovic and Krstic 1999, Pettersen and
Nijmeijer 2001).
The dirigible adapts itself and becomes a vector to be
taken into account in the future. A current use was in
Kosovo (2000) for the location of mines. For this pur-
pose, a UWB radar is integrated to the blimp. The
exact position of the detected mines is recorded for the
establishment of a chart of the mineelds. As the buoy-
ancy force provides an energy-free form of lift, blimps
are well suited for long-duration missions like intercon-
tinental transportation. Blimps are present in the
accesses of the large cities throughout the world.
Monitoring became a new mission where they excel.
Thus, with the Olympic games of Atlanta a blimp was
employed like an airborne platform monitoring the
Olympic site. In zones at risk, one can use the airship
to inspect the electric lines; dicult to achieve with
helicopters. All these examples are radio controlled.
Blimps referred to as UAVs have been used in
various contexts including some experimental work
(Coelho et al. 1998, Elfes et al. 1998, Zhang and
Ostrowski 1999, Hygounenc et al. 2000, Bestaoui and
Hamel 2000 and Beji et al. 2003). Their capability is con-
siderable in increasing manoeuvrability for tasks such as *Corresponding author. Email: beji@iup.univ-evry.fr
International Journal of Control
ISSN 00207179 print/ISSN 13665820 online 2005 Taylor & Francis Group Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/002071705000118643
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

transportation, surveillance and military applications
(De Paiva et al. 1999, Khoury and Gillet 1999).
In general, UAVs are not rigidly attached to a refer-
ence frame, so one cannot directly measure internal
conguration variables to generate a pose. Additionally,
the control inputs are exerted directly through the
application of forces and torques. The statement of this
problem was clearly posed in Zhang and Ostrowski
(1999) where a methodology is developed for incorpor-
ating the physical parameters of the blimp into the
image plane for performing vision-guided control.
In this case, tests are restricted to indoor areas.
In some studies such as Fossen (1994), Khoury and
Gillet (1999), motion is referenced to a system of ortho-
gonal body axes xed in the airship. The model used
was written originally for a buoyant underwater vehicle
(Fossen 1994). It was modied later to take into account
the specicity of the airship Khoury and Gillet (1999).
In this paper, we propose to control the model given
in Bestaoui and Hamel (2000). This dynamic model
has the particularity that the origin of the airship xed
frame is the centre of gravity, while in the cited works,
it is located in the centre of buoyancy. The centre of
buoyancy is the centre of the airship volume.
The design of robust controllers for aerial unmanned
vehicles requires the solving of many dicult problems
mainly due to the large variability of the models
during the dierent phases of ight (take o, cruise
ight, landing). In practice, the autopilots currently
used in the industry select the appropriate control law
depending on the phase of ight. In the case of an
airship with vectorized engines, the diculties increase
due to the existence of two radically dierent modes of
ight (aerostatic versus aerodynamic ight) involving
strong non-linearities in the model. Another specicity
is the diculty of clearly dening which degrees of
freedom are actuated. For example, in the aerodynamic
mode, the control surfaces become ecient only
when the air velocity is larger than a certain value.
Conversely, in the aerostatic mode, due to the vectoriza-
tion, the actuated directions of motion can change. In
some cases the airship behaves like an underactuated
system, while in other cases the means of actuation are
redundant.
In order to achieve these objectives, a robust guidance
and control system is required, capable of auto-piloting
and controlling the airship under an extremely wide
range of atmospheric and wind conditions. The success-
ful design of such a system rst requires an accurate
model of airship dynamics across its expansive ight
envelope, and a representative model of the expected
disturbances. The dynamics of an airship are markedly
dierent from traditional aircraft, with signicant
eects from added mass and inertia, and a much
higher sensitivity to wind. In this paper, a typical airship
conguration is rst sized to meet energy balance and
mass constraints. The geometry of this conguration is
then used to develop a general aerodynamic model for
the airship. The equations of motion with added mass
and inertia are developed.
In this paper we solve the tracking control feedback
problem for ascent and descent ight with only three
available control inputs: the main and tail thrusters
and the tilt angle of the propellers. The roll is totally
unactuated. The same input controls both pitch and
surge, while yaw and sway are related. The unactuated
dynamics implies constraints on acceleration. The
dynamic tracking control problem consists of nding a
feedback control law that asymptotically stabilizes the
relative equilibrium of the airship in ascent and descent
manoeuvre. The relative equilibrium in which the rate of
change (of magnitude) of the airships state vector is
zero and the result of the applied forces and moments
is constant which leads to helicoidal, circular or
straight-line trim trajectories.
This paper is organized as follows. The dynamic
model with kinematic resolutions are addressed in x 2.
The airships relative equilibrium problem is dened
and solved in x 3. In x 4, we develop the feedback track-
ing control objectives. More precisely, we consider the
helix path following which will be justied with respect
to the system input chattering and limits. Simulation
results with comments are the subject of x 5.
2. Airship structure and modelling
An overview of the system is given by gure 1. The
forces and moments are referred to a system of body-
xed axes, centred at the airship centre of gravity
(gure 2). We assume that the earth xed reference
frame is inertial, the gravitational eld is constant and
the airship is supposed to be a rigid body, meaning
that it is well inated. The aero-elastic eects are
ignored, the density of air is supposed to be uniform,
and the inuence of gust is considered as a continuous
disturbance, ignoring its stochastic character.
2.1. Kinematic descriptions
Two reference frames are considered in the derivation
of the kinematics and dynamics equations of motion.
These are the Earth xed frame R
f
and the body xed
frame R
m
(gure 2). This formulation is used for under-
water vehicles (Fossen 1994). The origin C of R
m
coincides with the center of gravity of the vehicle.
Its axes x
c
, y
c
, z
c
are the principal axes of symmetry
when available. They must form a right handed
orthogonal normalized frame. The position and the
Tracking control of trim trajectories of a blimp 707
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

orientation of the vehicle C in R
f
can be respectively
described by
j
1

x
y
z
0
@
1
A
, j
2

[
0
@
1
A
1
with Roll, pitch and [ Yaw angles (aeronautical
angles). Let Roll, pitch and [ Yaw angles, which
denotes the Euler parameters which are chosen to param-
eterize SO(3). Let us introduce j j
1
, j
2

T
where
j
1
x, y, z
T
is the position vector of the airship,
and V as the linear velocity of the origin and as the
angular velocity (expressed in the body xed frame)
V
u
v
w
0
@
1
A
,
p
q
r
0
@
1
A
. 2
The kinematics of the airship can be expressed in the
following way
_ j
1
j
1
_ j
2
j
2

Rj
2
0
0 Jj
2

v, 3
where v V,
T
. The orientation matrices R and J are
as Pettersen (1999)
R
c
[
c

s
[
c

c
[
s

s
[
s

c
[
s

s
[
s

c
[
c

s
[
s

c
[
s

s
[
s

0
B
@
1
C
A
Jj
2

1
c

1 s

0 c
[
c

0 s

0
B
@
1
C
A.
The following shorthand notation for trigono-
metric function is used: c
[
: cos[, s
[
: sin[,
t
[
: tan[, etc.
R 2 SO
3
denote the orthogonal rotation matrix that
species the orientation of the airship frame relative to
the inertial reference frame in inertial reference frame
coordinates. This description is valid in the region
,2 - - ,2.
Figure 1. The Understudy LSCAS200 airship.
Stern thruster
Earth frame
Lateral thrusters
Ailerons
Security valve
Hull
Blimp's frame
o x1
z
y
F1
F2
F3
B1
G
x
Gondola
(Yaw)
(Roll)
(Pitch)

y1
z1
Figure 2. Reference frames attached to the blimp.
708 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

2.2. Dynamic of the airship
In this paper, the dynamics model is dened as the set
of ordinary dierential equations relying the situation
of the vehicle in its position, velocity and acceleration
to the control vector. The translational part is separated
from the rotational part. As the blimp displays a very
large volume, its virtual mass and inertia properties
cannot be neglected. The dynamic model is expressed
in the airship xed frame, centred at the airship centre
of gravity as
_ j
1
j
1
Rj
2
V
_ j
2
j
2
Jj
2

)
4
M
_
VV MV D
V
V mg BR
T
e
z
F
1
F
2
I
_
I V MV D

R
T
e
z
BG

B
F
1
P
1
G F
2
P
2
G
9
>
>
=
>
>
;
5
m: mass of the airship, the propellers and
the actuators,
M: this 3 3 mass matrix includes both
the airships actual mass as well as the
virtual mass elements associated with
the dynamics of buoyant vehicles,
I : this 3 3 inertia matrix includes both
the airships actual inertia as well as
the virtual inertia elements associated
with the dynamics of buoyant vehicles,
with respect to G,
D
V
: a 3 3 aerodynamic forces diagonal
matrix,
D

: a 3 3 aerodynamic moments diago-

nal matrix,
F
1
and F
2
: Vectors of the propulsion forces,
e
z
(0 0 1)
T
: a unit vector,
B ,g : represents the magnitude of the buoy-
ancy force. is the volume of the envel-
ope, , the dierence between the density
of the ambient atmosphere ,
air
and
the density of the helium ,
helium
in
the envelope, g is the constant gravity
acceleration.
The symbol x denotes the vector cross product.
P
i
G: represents the position of the ith propel-
ler with respect of the centre of mass.
Terms MV and I show the centrifugal and
Coriolis components.
Added mass should be understood as pressure
induced forces and moments due to a forced harmonic
motion of the body which are proportional to the accel-
eration of the body. In order to allow the vehicle to pass
through the air, the uid must move aside and then close
behind the vehicle. As a consequence, the uid passage
possesses kinetic that it would lack if the vehicle was
not in motion. The mass of the dirigible is assumed to
be concentrated in the center of gravity. Then if we
consider the plane XZ as a plane of symmetry, the
mass and inertia matrices can be written as
M
m X
x
0 X
z
0 m Y
y
0
Z
x
0 m Z
z
0
@
1
A
, 6
where X
x
, Y
y
and Z
z
are the virtual mass terms of X, Y
and Z axes respectively.
I
I
xx
K
x
0 I
xz
K
z
0 I
yy
M
y
0
I
xz
N
x
0 I
zz
N
z
0
@
1
A
, 7
where I
xx
, I
yy
and I
zz
are inertias about the blimpprin-
cipal axes of inertia. The elements K
x
, M
y
and N
z
are
the virtual inertia terms of X, Y, Z about GX, GY and
GZ axes, respectively.
The mass and inertia matrices are positive denite.
We will assume that the added mass coecients are
constant. They can be estimated from the inertia ratios
and the airship weight and dimension parameters.
The aerodynamic force can be resolved into two
component forces, one parallel and the other perpendi-
cular to the direction of motion
D
V
diagX
u
, Y
v
, Z
w

diagL
p
, M
q
, N
r
.
The gravitational force vector is given by the dierence
between the airship weight and the buoyancy acting
upwards on it: mg BR
T
e
z
and the gravitational and
buoyant moments are given by: B R
T
e
z
BG

where
BG x
B
, y
B
, z
B
represents the position of the centre
of buoyancy with respect to the airship-xed frame.
The term buoyancy is used in hydrodynamics while
the term static lift is used in aerodynamics. The centre
of buoyancy is the centre of gravity of the displaced
uid. It is the point through which the static lift acts.
The centre of gravity is the point through which the
weight of the object is acting. The relationship between
the centre of gravity and the centre of buoyancy is an
important parameter. For the airship to remain stati-
cally level (aerodynamics and thrust eects are ignored
here), the centre of gravity should be directly below
the centre of buoyancy. If the centre of gravity sits
below the centre of buoyancy, then BG 0 0 z
B

T
.
Any horizontal oset will result in the airship adopting
a pitch angle. The vertical separation between these two
centres aects the handling characteristics of the airship.
Tracking control of trim trajectories of a blimp 709
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

Actuators provide the means for maneuvering the air-
ship along its course. An airship is propelled by thrust.
Propellers are designed to exert thrust to drive the
airship forward.
An airship is an under-actuated system with two types
of control: forces generated by thrusters and angular
inputs controlling the direction of the thrusters (j is
the tilt angle of the main propeller)
F
1

T
m
sin j
0
T
m
cos j
0
@
1
A
, F
2

0
T
t
0
0
@
1
A
,
where T
m
and T
t
represent respectively the main and tail
thrusters.
Thus in building the non linear six degrees of
freedom mathematical model, the additional following
assumptions are made
P
1
G
0
0
P
3
1
0
@
1
A
, P
2
G
P
1
2
0
0
0
@
1
A
.
In order to understand the action of dierent inputs on
the blimp, we need to develop the model given in (45).
As we will see in the following the input t
3
act only on
the dynamic of _ oo. However, the inputs t
2
and t
1
should control the rest of dynamics. Explicitly, the
dynamics of the blimp can be written in the form pre-
sented below: the six relations related to the dynamic
part (5) are subdivided onto to linear accelerations
_ uu
1
m
11
X
u
u m
22
rv m
33
qw

1
m
11
B mgs

1
m
11
t
1
_ vv
1
m
22
Y
v
v m
33
pw m
11
ru

1
m
22
B mgc

1
m
22
t
2
_ ww
1
m
33
Z
w
w m
11
qu m
22
pv

1
m
33
B mgc

1
m
33
t
3
_ pp
1

n
I
33
L
p
p I
13
N
r
r
I
2
33
I
2
13
I
33
I
22
qr I
13
I
11
I
22
I
33
pq
I
13
X
u
Y
v
uw I
33
Z
z
Y
v
vw
o

I
33
Bz
b
c

I
13
P
1
2
t
2
_ qq
1
I
22
n
M
q
q I
13
p
2
I
13
r
2
I
11
I
33
pr X
u
Z
z
uw
o

1
I
22
Bz
b
s

1
I
22
P
3
1
t
1
_ rr
1

n
I
13
L
p
p I
11
N
r
r
I
2
13
I
2
11
I
11
I
22
pq I
13
I
11
I
22
I
33
qr
I
11
Y
v
X
u
uw I
13
Y
v
Z
z
vw
o

I
13
Bz
b
c

I
11
P
1
2
t
2
8
connected to the six kinematic equations, which are
obtained from equation (4)
_ xx c
[
c

u s

v s

w s
[
c

v s

w
_ yy s
[
c

u s

v s

w c
[
c

v s

w
_ zz s

u s

v c

w
_
p t

q c

r
_
c

q s

r
_
[[
1
c
s

q r 9
where I
2
13
I
11
I
22
and t
1
, t
2
and t
3
are the three
inputs of the system such that
t
1
t
2
t
3
0
@
1
A

T
m
sinj
T
t
T
m
cosj
0
@
1
A
10
Remark 1: A strategy to design a stabilizing controller
of position and attitude of systems (1011) was pro-
posed in Beji (2003). We have proved that the airship
cannot be stabilized to a point using continuous pure-
state feedback law. The problem was solved using the
averaging approach with an explicit homogeneous
time-varying feedback law.
3. Relative equilibrium: airship trimmed
ight conditions
The fundamentals of ight are in general: straight and
level ight (maintenance of selected altitude), ascents
and descents, level turns, wind drift correction and
ground reference manoeuvre. Trim is concerned with
the ability to maintain ight equilibrium with controls
xed. A trimmed ight manoeuvre is characterized by
the rate of change (of magnitude) of the airships state
vector (in the body-xed frame) and the resultant
of the applied forces and moments is xed. In a trimmed
manoeuvre, the airship will be accelerating under
710 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

the action of non-zero resultant aerodynamic and
gravitational forces and moments. These eects will be
balanced by eects such as centrifugal and gyroscopic
inertial forces and moments. The trim problem is gener-
ally formulated as a set of non-linear algebraic equations
_ uu _ vv _ ww _ pp _ qq _ rr 0. 11
Using (11), the angular velocity can be rewritten as
p
_

_
[[s

q
_
c

_
[[s

r
_
s

_
[[c

.
Dierentiating versus time and nullifying these
derivatives, we obtain
p
0

_
[[
0
s

0
q
0

_
[[
0
c

0
s

0
r
0

_
[[
0
c

0
c

0
with one of the solutions given by
_
0
_
0
_
[[
_
[[
0
.
9
>
>
=
>
>
;
12
Thus

0

0
[
_
[[
0
t.
9
>
=
>
;
. 13

0
,
0
, and
_
[[
0
are constants. Using (11), trimming
trajectories are characterized by
_ xx a
x
cos
_
[[
0
t b
x
sin
_
[[
0
t
_ yy a
y
cos
_
[[
0
t b
y
sin
_
[[
0
t
_ zz _ zz
0
sin
0
u
0
cos
0
sin
0
v
0
cos
0
cos
0
w
0
where
a
x
u
0
cos
0
v
0
sin
0
sin
0
w
0
cos
0
sin
0
b
y
a
x
b
x
v
0
cos
0
w
0
sin
0
a
y
b
x
.
9
>
>
>
=
>
>
>
;
14
Integrating, we obtain
rs
xs
ys
zs
0
B
@
1
C
A 15
with
x
a
x
_
[[
0
sin
_
[[
0
V
e
s

b
x
_
[[
0
cos
_
[[
0
V
e
s

y
b
x
_
[[
0
sin
_
[[
0
V
e
s

a
x
_
[[
0
cos
_
[[
0
V
e
s

z
_ zz
0
V
e
s,
9
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
;
16
where s represents the curvilinear abscissa and we
suppose a uniform motion such that
s V
e
t t

u
2
0
v
2
0
w
2
0
.
q
17
The trajectories represented by these equations are
vertical axis helices with constant curvature and torsion.
The most general trim condition resembles a spin mode.
The spin axis is always directed vertically in the trim. The
trim condition can be a turning (about the vertical axis),
descending or climbing (assuming constant air density
and temperature), side-slipping manoeuvre at constant
speed. More conventional ight conditions such as
hover, cruise, auto-rotation or sustained turns are also
trims.
4. Tracking control of trimmed ights
In this section we develop a feedback law that stabilizes
the tracking error dynamics: the dierence between
the desired state dynamic and the true state one.
The reference trajectory (trim trajectory) is subject of
_ uu
r
_ vv
r
_ ww
r
_ pp
r
_ qq
r
_ rr
r

r
0
and
_
[[
r

_
[[
r
0
cte. Helix for motion planning has
adopted to perform ascent and descent ights. This is
in order to reach a desired altitude.
As we can see, the dynamic model of the blimp
presents a complex structure with height degree of
coupling. This complexity is due to the local representa-
tion of the model (gyroscopic terms). Thus, and in order
to adapt these equations to the tracking feedback task,
we propose a coordinate changes for velocities which
is a global dieomorphism transformation
U c

u s

v s

w
V c

v s

w
W s

u c

v c

w
9
>
=
>
;
18
Tracking control of trim trajectories of a blimp 711
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

and let us consider
z
1
c
[
x s
[
y
z
2
s
[
x c
[
y
z
3
z.
9
>
=
>
;
19
Further we consider the following local transformation
to attitude velocities, with 2,2, ,2
Q
1
c

q c

r
R c

q s

r
P p.
9
>
>
>
=
>
>
>
;
20
Then, the state equations of the UAV, in the new coor-
dinates U, V, W, P, Q, R, z
1
, z
2
, z
3
, , , [, take the
form
_
UU RW P s

Qs

V c

_ uu s

_ vv s

_ ww
_
VV P s

Qs

u c

w c_ vv s

_ ww
_
WW RU P s

Qc

v s

_ uu c

_ vv c

_ ww
_
PP _ pp
_
QQ
1
c

RP 2Qs

_ qq c

_ rr

_
RR c

P s

QQ c

_ qq s

_ rr
_ z
1
z
1
Qz
2
U
_ z
2
z
2
Qz
1
V
_ z
3
z
3
W
9
>
=
>
;
21
_
P s

Q
_
R
_
[[ Q.
9
>
>
=
>
>
;
22
Consider conditions of trimmed ight, the reference
trajectory in the new system coordinates is given by
_
U
r
U
r
0
_
V
r
V
r
0
_
W
r
W
r
0
_
P
r
P
r
0
_
Q
r
Q
r
0
_
z
r
1
z
r
1
Q
r
z
r
2
U
r
_
z
r
2
z
r
2
Q
r
z
r
1
V
r
_
z
r
3
z
r
3
W
r
_

r
0
_

r
0
_
[
r
[
r

_
[
r
0
. [
r
0
.
9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;
23
We dene an error variable by e

i

i

r
i
and its time
derivative _ ee

i
_
i
_
r
i
. The dynamic of errors in the
new system coordinates is as
_ ee
U
RW P s

Qs

V c

_ uu s

_ vv s

_ ww
_ ee
V
P s

Qs

U c

W c_ vv s

_ ww
_ ee
W
RU P s

Qc

v s

_ uu c

_ vv c

_ ww
_ ee
P

_
PP
_ ee
Q

1
c

RP 2s

Q s

_ qq c

_ rr
_ ee
R
c

P s

QQ c

_ qq s

_ rr
_ ee
z
1
Q
r
e
z
2
z
2
e
Q
e
U
_ ee
z
2
Q
r
e
z
1
z
1
e
Q
e
V
_ ee
z
3
e
W
_ ee

P s

Q
_ ee

e
R
_ ee
[
e
Q.
9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;
24
Notice that the system above is controlled by t
1
, t
2
and
t
3
but through _ uu, _ vv, _ ww, _ pp, _ qq and _ rr. In order to make
appear the control inputs, let us consider the proposition
below.
Proposition 1: With the proposed feedback controllers
t
1
X
u
u m
22
rv m
33
qw B mgs

m
11
t
u
z
R
I
22
c

P
3
1
R z

I
22
c

P
3
1
e

t
2
m
33
pw m
11
ru B mgc

m
22
t
v
z
p
I
33
L
p
I
13
P
1
2
p
r
z
Q
c

I
11
P
1
2
e
Q

N
r
P
1
2
r
z
[
c

I
11
P
1
2
e
[
z

I
13
P
1
2
e

t
3
Z
w
w m
11
qu m
22
pv B mgc

m
33
t
w
9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;
25
the following dynamics are obtained
_ uu ~ tt
u
_ vv ~ tt
v
_ ww ~ tt
w
9
>
=
>
;
26
712 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

with
~ tt
u
t
u
z
R
I
22
m
11
c

P
3
1
R
z

I
22
m
11
c

P
3
1
e

~ tt
v
t
v

1
m
22
Y
v
v z
p
I
33
L
p
m
22
I
13
P
1
2
p
r
z
Q

m
22
c

I
11
P
1
2
e
Q

N
r
m
22
P
1
2
r
z
[

m
22
c

I
11
P
1
2
e
[
z

m
22
I
13
P
1
2
e

~ tt
w
t
w
.
9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;
27
t
u
, t
v
and t
w
will be specied later, and the extra terms
in the control inputs are adapted to the dynamics of
_
PP,
_
QQ and
_
RR.
Now, we transform the rst three equations in the
systems dynamics which are deduced from (28) and
(30), as follows
_ ee
U
RW P s

Qs

V c

~ tt
u
s

~ tt
v
s

~ tt
w
_ ee
V
P s

Qs

U c

W c~ tt
v
s

~ tt
w
_ ee
W
RU P s

Qc

v s

~ tt
u
c

~ tt
v
c

~ tt
w
9
>
>
=
>
>
;
28
The proper choice of t
U
, t
V
and t
W
, such that the
following satises
~ tt
u
~ tt
v
~ tt
w
0
B
B
@
1
C
C
A

0 c s

0
B
B
@
1
C
C
A
1

t
U
RW P s

Qs

V
t
V
P s

Qs

U c

W
t
W
RU P s

Qc

v
0
B
B
@
1
C
C
A
29
allows to write the following dynamics with inputs t
U
,
t
V
and t
W
_ ee
U
t
U
_ ee
V
t
V
_ ee
W
t
W
.
Lemma 1: Under the tracking controller given by (29),
in closed loop, the attitude error dynamics behavior of
the blimp take the following stable linear approximation
_ ee
P
z
p
I
33
L
p

e
p
z

_ ee
Q
z
Q
e
Q
z
[
e
[
_ ee
R
z
R
e
R
z

,
9
>
>
=
>
>
;
30
where the blimps characteristics are constant and verify
I
33
> 0, L
p
- 0 and <0 as I
2
13
- I
11
I
22
.
Proof: Recall that _ ee
P
_ pp, one incorporating the input
t
2
given by (29) and one taking the linear approximation
of this system, we can easily verify the given error in
closed loop. We choose to control the dynamic of e
Q
through _ rr which depend on t
2
. So, one introducing t
2
via _ rr (10) in _ ee
Q

_
QQ (25), after eliminating the nonlinear
terms, we get easily _ ee
Q
z
Q
e
Q
z
[
e
[
. As soon as for
e
R
, it will be controlled in closed loop by _ qq, consequently
by t
1
. The closed loop form of this variable can be easily
veried. Note that the pith angle ,2 is a singular-
ity, this attitude cannot be reached by the blimp.
Let us point out that any conclusion on the tracking
problem could integrate all interactions in the system
dynamic, including kinematic relations. This is the
subject of the following subsection.
4.1. Kinodynamics feedback stabilization
The tracking control problem investigated above is
reduced to the stabilization of the following blimps
error dynamics
_ ee
U
t
U
_ ee
V
t
V
_ ee
W
t
W
_ ee
P
z
p
I
33
L
p

e
p
z

_ ee
Q
z
Q
e
Q
z
[
e
[
_ ee
R
z
R
e
R
z

_ ee
z
1
Q
r
e
z
2
z
2
e
Q
e
U
_ ee
z
2
Q
r
e
z
1
z
1
e
Q
e
V
_ ee
z
3
e
W
_ ee

e
P
_ ee

e
R
_ ee
[
e
Q
.
9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;
31
The stabilization control problem we will discuss
thus consists of nding an appropriate state feed-
back controller for the trimmed ight into the new
Tracking control of trim trajectories of a blimp 713
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

system errors. The controllers are t
U
, t
V
and t
W
such
that the error dynamics presented above are asymptoti-
cally (exponentially) stable about the origin 0
R
12 .
In the rst step of the development of the stabilizing
controller, we decompose the system (35) in two uncon-
nected sub-systems. The rst one is controlled by t
W
, as
_ ee
W
t
W
_ ee
P
z
p
I
33
L
p

e
P
z

_ ee
R
z
R
e
R
z

_ ee
z
3
e
W
_ ee

e
P
_ ee

e
R
9
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
;
32
and the second system is controlled by t
U
and t
V
, as
_ ee
U
t
U
_ ee
V
t
V
_ ee
Q
z
Q
e
Q
z
[
e
[
_ ee
z
1
Q
r
e
z
2
z
2
e
Q
e
U
_ ee
z
2
Q
r
e
z
1
z
1
e
Q
e
V
_ ee
[
e
Q
9
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
;
33
Subsystems (36) and (37) present some dynamics which
are in closed loop, then it remains only the analysis of
their stabilities. While to solve the control objectives,
let us rst announce the result concerning the stability
of cascaded systems (Coron 1998).
Theorem 1: We consider the system with the control
input u(x)
_ xx f x, ux 34
and f x, ux: IR
n
IR
p
! IR
n
is a regular function.
Assume that the origin 0
IR
n of this system is C
1
asympto-
tically stable. Then, for positive and large enough values
of k
1
, k
2
, k
3
, k
4
, . . . , k
p
, the origin (x 0, y 0) of the
system
_ xx f x, y
_ yy
1
k
1
y
1
u
1
x
_ yy
2
k
2
y
2
u
2
x
.
.
.
_ yy
p
k
p
y
p
u
p
x
9
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
;
35
is C
0
asymptotically stable.
The asymptotic (exponential) stability of a part of
subsystem (36) is given by the following lemma.
Lemma 2: Under the continuous feedback controller
t
W
k
W
e
W
k
z
3
e
z
3
36
and z
R
, z

, z

are positive dene and stable polynomial

coecients, with k
W
>0 and k
z
3
> 0, e
z
3
, e

and e

in
(36) are exponentially (asymptotically) stable errors.
Proof: From (36), we get
_ ee
W
t
W
_ ee
z
3
e
W
,
)
37
where it is straightforward to ensure the exponential
stability of e
z
3
with e
W
k
z
3
e
z
3
. Further by virtue of
theorem 1, we construct t
W
as given in (40). Now, we
consider the remaining following dynamics, which are
in closed loop
_ ee
P

I
33
L
p

e
P
z

_ ee

e
P
9
=
;
38
_ ee
R
z
R
e
R
z

_ ee

e
R
)
39
The second time derivative below, can be computed.
We obtain ee

z
p
I
33
L
p
, _ ee

. As soon as
for ee

z
R
_ ee

, then the asymptotic stability of

e

, e

and their time derivatives can be asserted by the

identifying of these derivative to a stable polynomial
form. Moreover, e
p
and e
R
converge exponentially
to zero.
Now, we will interest to the subsystem-2. From
theorem 1, it is sucient to show that the following
reduced subsystem, obtained from subsystem-2, by
taking e
U
u
1
and e
V
u
2
as a new control variables,
_ ee
Q
z
Q
e
Q
z
[
e
[
_ ee
z
1
Q
r
e
z
2
z
2
e
Q
u
1
_ ee
z
2
Q
r
e
z
1
z
1
e
Q
u
2
_ ee
[
e
Q
9
>
>
>
=
>
>
>
;
40
is asymptotically stable. The stability result is given by
the following lemma.
Lemma 3: Consider the functions
u
1
k
z
1
e
z
1
Q
r
e
z
2
z
2
e
Q
u
2
k
z
2
e
z
2
Q
r
e
z
1
z
1
e
Q
)
41
714 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

and the feedback controller
t
U
k
1
e
U
u
1

t
V
k
2
e
V
u
2

)
42
and take z
Q
, z
[
as coecients of a stable polynomial form,
with k
1
> 0, k
2
> 0, then the errors e
z
1
and e
z
2
are expo-
nentially stable. Moreover, e
[
is asymptotically stable.
Proof: Introduce (45) into (44), we can easily verify that
_ ee
z
1
k
z
1
e
z
1
and _ ee
z
2
k
z
2
e
z
2
. Moreover, the second
time derivative of e
[
leads to ee
[
z
Q
_ ee
[
z
[
e
[
. Then
the well choose of z
Q
and z
[
guarantees the asymptotic
stability of the error in attitude e
[
. The end of the
proof is asserted by the theorem 1, which permits to
conclude on the form given in (46).
Remark 2: By classical stability results, the local sta-
bility of the initial tracking error system (28) is obtained
by the stability of the linearized system one (35).
5. Simulation results
The lighter than air platform used for simulations is the
AS200 by Airspeed Airships. It is a remotely piloted air-
ship designed for remote sensing. It is a nonrigid 6 m
long, 1.4 m diameter and 7.6 m
3
volume airship. In this
section, we present some simulation results for the
ascent and descent ight. The trimmed ight is charac-
terized by a vertical axis helix with constant curvature.
More details about the trimmed ight and the helix
characteristics are given in Bestaoui and Hima (2001).
The blimps parameters are as follows (International
System Units):
.
blimps total masse is m 9.07, with the nacelle mass
m
n
1.58.
.
added masses X
x
1.13, Y
y
7.25, Z
z
7.25,
K
x
0, M
y
8.87, N
z
8.87.
. inertial parameters around the principal axes of iner-
tia: I
xx
2.19, I
yy
18.85, I
zz
18.76 and I
xz
0.
. inertial terms: I
11
I
xx
K
x
2.19, I
22
I
yy
M
y

27.73, I
33
I
zz
N
z
27.63 and I
13
I
xz
K
x
0.22.
.
term: I
2
13
I
11
I
22
60.89.
.
positions of input forces F
1
and F
2
: P
3
1
1 and
P
1
2
3.
. aerodynamic coecients: X
u
Y
v
Z
w
L
p
N
q

M
r
10.
. buoyancy and gravity magnitudes: mg 89 and
B 72.2 where ,
a
1.3(air) and ,
h
0.3(helium).
All the tracking results are tested with a simulator
performed with the Matlab environment. The following
gures give an idea about the trimmed path following
with the proposed feedback tracking controller.
Further, we have evaluated the six attitude/position
tracking error variables and the magnitude of the three
blimps inputs. Helices are characterized and adopted
for ascent and descent ight manoeuvre. It will be con-
sidered as motion planning to reach a given altitude with
constant inputs. Tests are subject of conditions given
above. The results are sketched in gure 3 for the
descent ight and gure 10 for the ascent ight. The
tracking errors in attitude and position are given by
gure 11. The last, displays a good behaviour of the
blimp in descent ight even if the start conguration of
the blimp do not belong to the helix. The ascent ight
manoeuvre is demonstrated in gure 10. Adopting
an helix to the blimp motion planning, permits to
respect the magnitude limit of the actuators (we can
see gures 48).
As it has been justied above, a straight line (vertical
or inclined) veries the trim ight conditions. While,
this trajectory cannot be followed by the blimp in its
ascent/descent ight manoeuvre (gures 1819).
More precisely, we reach the pose in altitude with an
helix in the vicinity of the line. In addition, the values
30
20
10
0
10
20
30
40
20
0
20
40
250
200
150
100
50
0
50
x,x
r
(m)
Descent trimmed flight
y,y
r
(m)
z
,
z
r
(
m
)
Figure 3. Descent tracking with helix like a trimmed ight.
0 2 4 6 8 10 12 14 16 18 20
800
700
600
500
400
300
200
100
0
100

1
:
T
m
s
i
n
(

)
(
N
)
times(s)
Figure 4. Feedback tracking controller t
1
.
Tracking control of trim trajectories of a blimp 715
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

0 5 10 15 20
0
10
20
e
x
(
m
)
Different tracking errors of the blimp: trimmed flight
0 5 10 15 20
0
5
10
15
e
y
(
m
)
0 5 10 15 20
0
5
10
times(s)
e
z
(
m
)
0 5 10 15 20
15
10
5
15
10
5
15
10
5
e

(
d
e
g
)
0 5 10 15 20
e

(
d
e
g
)
0 5 10 15 20
times(s)
e

(
d
e
g
)
Figure 9. Sketch of the descent tracking errors.
0 5 10 15 20 25
18
16
14
12
10
8
6
4
2
0
2

(
d
e
g
)
times(s)
Figure 8. Resulting feedback tracking controller j.
0 2 4 6 8 10 12 14 16 18 20
350
300
250
200
150
100
50
0
50
100
150

3
:
T
m
c
o
s
(

)
(
N
)
times(s)
Figure 6. Feedback tracking controller t
3
.
0 2 4 6 8 10 12 14 16 18 20
1000
500
0
500

2
:
T
t
(
N
)
times(s)
Figure 5. Feedback tracking controller t
2
.
0 5 10 15 20 25
400
200
0
200
400
600
800
1000
i
n
p
u
t

T
m
(
N
)

times(s)
Figure 7. Resulting feedback tracking controller T
m
.
716 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

of inputs are important in this case, with large errors
in x and y directions.
In the case of a circler path, which is a trim trajectory
for the blimp (obtained for z
r
cst, our tracking
tests are tested and showed in gure 12. Initially the
blimp was positioned at x0, y 0, z 2 m, and the
desired circle was placed at altitude z 200 m.
In this case, the derived control law involves important
values for t
1
, t
2
and t
3
(gures 1315). Moreover
j-control shows a pick which is equal to 35

(gure
16). Hence, important and undesirable values in inputs
provided by tracking of a circle at a given altitude
without motion planning. These amplitudes may
exceed the limit of motors. Diculties are demonstrated
when tracking is subject of an inclined line (gure 19).
This can be explained by the fact that a blimp cannot
follow a straight line with [-attitude maintained
constant.
In conclusion, following a straight-line or a circle
path by the blimp, with control magnitude limita-
tions and without chattering can be ensured by the
proposed controller. To do this, it is necessary to
generate a helix which permits to reach the desired
altitude with inputs constant in magnitudes.
Consequently at this altitude, the blimp can follow a
xed circle or a desired straight-line where the chattering
behaviour can be cancelled and the actuator limits
can be respected. The question which is actually out
0 5 10 15 20
0
10
20
e
x
(
m
)
Different tracking errors of the blimp: trimmed flight
0 5 10 15 20
0
5
10
15
e
y
(
m
)
0 5 10 15 20
0
5
10
times(s)
e
z
(
m
)
0 5 10 15 20
15
10
5
15
10
5
15
10
5
e

(
d
e
g
)
0 5 10 15 20
e

(
d
e
g
)
0 5 10 15 20
times(s)
e

(
d
e
g
)
Figure 11. Sketch of the ascent tracking errors.
30
20
10
0
10
20
30
40
20
0
20
40
50
0
50
100
150
200
250
x,x
r
(m)
Descent trimmed flight
y,y
r
(m)
z
,
z
r
(
m
)
Figure 10. Helix-path for ascending ight.
30
20
10
0
10
20
30
40
20
0
20
40
50
0
50
100
150
200
250
x,x
r
(m)
Descent trimmed flight
y,y
r
(m)
z
,
z
r
(
m
)
Figure 12. Circular path following like a trimming ight.
Tracking control of trim trajectories of a blimp 717
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

of the area of this paper is: how can us bring the system
between the helix, circle and the straight-line paths
taking into account the chattering behaviour of inputs
and the actuator limits?
6. Conclusion
The complexity in coupling of the blimps dynamic
equations were transformed in adequate form in this
15
10
5
0
5
10
15
15
10
5
0
5
10
15
250
200
150
100
50
0
50
x,x
r
(m)
Tracking of the trimmed flight
y,y
r
(m)
z
,
z
r
(
m
)
Figure 18. Tracking of a vertical straight line.
0 5 10 15 20 25
300
200
100
0
100
200
300
400
500
600
700
i
n
p
u
t

T
m
(
N
)

times(s)
Figure 17. Input T
m
behaviour to reach the circular-path.
0 2 4 6 8 10 12 14 16 18 20
1800
1600
1400
1200
1000
800
600
400
200
0
200

2
:
T
t
(
N
)
times(s)
Figure 14. Input t
2
behaviour to reach the circular-path.
0 2 4 6 8 10 12 14 16 18 20
400
300
200
100
0
100
200
300

3
:
T
m
c
o
s
(

)
(
N
)
times(s)
Figure 15. Input t
3
behaviour to reach the circular-path.
0 5 10 15 20 25
20
10
0
10
20
30
40

(
d
e
g
)
times(s)
Figure 16. Input j behaviour to reach the circular-path.
0 2 4 6 8 10 12 14 16 18 20
0
100
200
300
400
500

1
:
T
m
s
i
n
(

)
(
N
)
times(s)
Figure 13. Input t
1
behaviour to reach the circular-path.
718 L. Beji and A. Abichou
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2

paper. The feedback tracking control problem was
clearly posed in the new systems coordinates and
solved for trimmed ight manoeuvre. Helices like Path
following for high system desired altitude, reduce the
chattering behaviour in inputs and respect the actuator
limits. Moreover, inputs can be maintained constant
during the ight. We have shown that a circle/
straight-line like trimming ight with a xed altitude,
which can be adopted for camera and TV platforms or
for specialized scientic tasks, cannot be reached
directly. Otherwise, the system involves chattering in
inputs and increasing in control magnitudes which
cannot be supported by the actuators. Our next problem
consists to test the exposed results and to nd a method
that permits to bring smoothly the blimp between
helices, circles and straight lines.
Acknowledgements
This work was performed while the rst author was vis-
iting the Tunisian Polytechnic School (LIM) supported
by the SERST programm. He gratefully acknowledges
this support.
References
E. Altug, J. Ostrowski and R. Mahony, Control of a quadrotor
helicopter using visual feedback, Proceedings of the IEEE
Conference on Robotics and Automation, Washington DC,
Virginia, USA, 2002.
L. Beji, A. Abichou and Y. Bestaoui, Position and attitude control
of an undeactuated autonomous airship, International Journal
of Dierential Equations, 8, pp. 231255, 2003.
Y. Bestaoui and T. Hamel, Dynamic modeling of small autonomous
blimps, Proceedings of the Conference on Methods and Models
in Automation and Robotics, Miedzyzdroje, Poland, 2000,
pp. 579584.
Y. Bestaoui and S. Hima, Trajectory of a dirigeable in high constant
altitude flight, Proceedings of the Conference IFAC Nonlinear
Control Systems, Saint Petersburg, Russia, 2001.
D.M. Boskovic and M. Krstic, Global attitude/position regulation
for underwater vehicles, International Journal of Systems Science,
309, pp. 939946, 1999.
L. Coelho, M. Campos and V. Kumar, Computer vision-based
navigation for autonomous blimps, International Symposium on
Computer Graphics, Image Processing and Vision, Rio de Janeiro,
Brazil, November 1998, pp. 287294.
J.M. Coron, On the stabilization of some nonlinear control systems:
results, tools, and applications, Technical Report, NATO
Advanced Study Institute, Montreal, July 27August 7.
E.C. de Paiva, S.S. Bueno and S.B.V. Gomes, A control system
development environment for AURORAs semi-autonomous
robotic airship, Proceedings of the IEEE Conference
on Robotics Automation, Detroit. Michigan, USA, May 1015
1999, pp. 23282335.
A. Elfes, S.S. Bueno, M. Bergerman and J.J.G. Ramos, A semi-
autonomous robotic airship for environmental monitoring mission,
Proceedings of the IEEE Conference on Robotics and Automation,
PAYS, Leuven, Belgium, 1998, pp. 34493455.
I. Fantoni and R. Lozano, Nonlinear Control Mechanical
Underactuated Systems, London: Springer Verlag, 2001.
T.I. Fossen, Guidance and Control of Ocean Vehicles Chichester:
John Wiley & Sons Ltd, 1994.
T.R. Hamel, R. Mahony, R. Lozano and J. Ostrowski, Dynamic
modelling and con-figuration stabilization for an X4-flyer,
Proceedings of the International Federation of Automatic Control
Symposium, Barcelona, Spain, 2002.
J. Hauser, S. Sastry and G. Meyer, Nonlinear control design
for slightly nonminimum phase systems: application to V/STOL
aircraft, Automatica, 28, pp. 665679, 1992.
E. Hygounenc, P. Soures and S. Lacroix, Modelisation dun dirigeable:
etude cinematique et dynamique, Technical Report 426, Laas-cnrs,
Toulouse, France, 2000.
G.A. Khoury and J.D. Gillet (Eds), Airship Technology, London:
Cambridge University press, 1999.
L. Marconi, A. Isodori and A. Serrani, Autonomous vertical landing
on an oscillating platform: an internal-model based approach,
Automatica, 38, pp. 2132, 2002.
P. Pound, R. Mahony, P. Hynes and J. Roberts, Design of a
four rotor aerial robot, Proceedings of the Australasian
Conference on Robotics and Automation, Auckland, November
2729, 2002.
R.V. Treuren, Nine Counter Measures Airship to the Arctic
Symposium, Winnipeg, Manitoba, October 2224, 2000.
H. Zhang and J.P. Ostrowski, Visual servoing with dynamics:
control of unmanned blimp, Proceedings of the IEEE
Conference on Robotics and Automation, Detroit, Michigan, USA,
May 1015, 1999.
250
200
150
100
50
0
50
4
2
0
2
4
6
250
200
150
100
50
0
50
x,x
r
(m)
Tracking of the trimmed flight
y,y
r
(m)
z
,
z
r
(
m
)
Figure 19. Tracking of an inclined straight line.
Tracking control of trim trajectories of a blimp 719
D
o
w
n
l
o
a
d
e
d

b
y

[
I
s
t
a
n
b
u
l

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y
]

a
t

0
6
:
1
4

0
3

O
c
t
o
b
e
r

2
0
1
2