[ieee 2009 ieee international conference on control applications (cca) - st. petersburg, russia...
TRANSCRIPT
State Estimation for Integrated Longitudinal and Lateral Car
Following Control
Ahmed Chaibet, Moussa Boukhnifer and Cherif Larouci
Abstract— This paper discusses an integrated vehicle con-trol and vehicles states estimations. The non linear observer,based on sliding mode approach is presented for estimationof the vehicle states. The considered technique is appliedto the estimation problem for automated vehicles operatingas two vehicles following. The integrated control consists ofsecond order sliding mode, based on twisting algorithm. Thedeveloped controller is tested in scenario of automated driving.Both performance of observer and controller are presented todemonstrate the effectiveness of the sliding mode observer andsliding mode controller in various maneuvers.
Index Terms— Non linear observer, Sliding mode, Twistingalgorithm, Stop and Go, Vehicle dynamics.
NOMENCLATURE
vx/vy Longitudinal/lateral vehicle speeds
Tc Equivalent drive and brake torque
Trr Rolling resistance torque
Cx/Cy Longitudinal/lateral aerodynamic drag coefficients
m Vehicle mass
Ie f f /Iz
Effective longitudinal inertia/ inertia moment about
the yaw axis through the vehicle center of gravity
Fy f /Fyr Cornering forces at the front tires/rear tires
l f /lr
Distances of the front and rear tires
from vehicle’s center of gravity
α f /αr Slip angle of the front and rear tires
c f /cr Cornering stiffness of the front and rear tires
δ f Steering angle
I. INTRODUCTION
During the two past decades research in a problem of car-
following has attracted the attention of several governmen-
tal institutions with the association of research laboratories
such as in the USA, in Japan and in Europe are focused
respectively on PATH program, the Intelligent Transporta-
tion System program and Chauffeur Project where safety,
traffic congestion’s and the energy consumption are the
mainly topics.These works treat aspects controls lateral and
longitudinal as well in simulation, in experimental phases.
Even though, in the literature, several vehicle models have
been proposed, which take into account variations of critical
parameters such as speed, tire adhesion and which describe
complex phenomena concerning a vehicle behavior. In the
M. Boukhnifer, A. Chaibet and C. Larouci are with LaboratoireCommande et Systemes, ESTACA, 34, rue de Victor Hugo, Levallois-Perret, 92300, France, e-mail:((achaibet, mboukhnifer,
clarouci )@estaca.fr).
control design, the longitudinal mode and the lateral one are
usually treated in a separate way via suitable assumptions.
A design of an integral lateral and longitudinal by using
the real-time trajectory curvature information of the leader
vehicle for vehicle following. Both longitudinal and lateral
controller were designed to guarantee that the vehicle
trailed the leader by tacking on count of the trajectory of
the leader vehicle. The longitudinal controller maintained
a safety distance between the two vehicles and the lateral
controller minimized the alignment angles between the two
vehicles is proposed in [4]. A longitudinal autonomous and
cooperative controller is treated and implemented in [12].
A robust controller synthesis using a backstepping concept
design procedure for both longitudinal and lateral dynamic
of car passenger where performances and stability of the
suggested controller are finally highlighted by some studies
in simulations in various situations is proposed [2]. On the
automated vehicles concept, the vehicle must be able to
achieve a whole of autonomous functions. Among these
functions such as: the heading variation, change maneuver
and a vehicle follower control system for a single vehicle
by maintaining a safety distance [3]. To conclude these
functions, the vehicle is equipped with proprioceptive and
exteroceptive sensors. To measure the whole vehicle states
dynamic is inconceivable for different reasons as economic
and of technical feasibility. In order to overcame these
drawbacks, a solution is to act for the observer concept.
For example, [5] has proposed an approach which combines
an extended Kalman filter for the estimation of the vehicle
dynamics and a Luenberger observer for the estimation of
the road slope. In [6] the author has developed an observer
of the lateral velocity and an integrated control combining
steering and braking/traction in order to improve vehicle
dynamics performance and stability. In this paper a sliding
mode observer based on the algorithm of optimal control
for the second order non linear system is addressed. The
estimation of the vehicle states retains a complexity of the
model and strong non linearities. This observer estimates
the non accessible variables an despite external disturbances
like the variation of road adhesion ( dry, softened, snow
covered, icy...). Sliding mode observer is known to be
robust technique that is an appropriate for the estimating
uncertain systems. High robustness is maintained against of
the various of uncertainties such as external disturbances
and measurement error. This nonlinear observer will be
used to estimate the leader and follower vehicle state’s.
Our emphasis within the development of this observer is
18th IEEE International Conference on Control ApplicationsPart of 2009 IEEE Multi-conference on Systems and ControlSaint Petersburg, Russia, July 8-10, 2009
978-1-4244-4602-5/09/$25.00 ©2009 IEEE 1140
to build the non measurable states and it will be used on
the control law synthesis. The problem considered, in this
paper is the design of an integrated lateral and longitudinal
controller that enables the follower to track the leader
vehicles trajectory by generating suitable steering angle δ f ,
while maintaining a desired following distance by through
of composite torque Tc, is carried out according to the sign
by a engine torque or a brake torque by using the states
estimation.The present paper is organized as follows: in
section 2 a vehicle modeling and positioning are addressed.
In the section 3, the description and the synthesis of the
observer design is given. Section 4 deals with the synthesis
of longitudinal and lateral controllers based on the second
order sliding mode. While the validation of this technique
by some simulation examples of scenarios of a driving is
devoted on the section 5. Conclusion and perspective of
this work are presented in the section 6.
II. VEHICLE MODELING AND POSITIONING
Fig. 1. Relative positioning of vehicles
One gives in the section a simplified three degree-of-
freedom vehicle model has been used for designing the
control strategy in this study. Under the assumptions that
there is no longitudinal slip between the tire and the road,
the pitch, the roll and the vertical dynamics are neglected.
The three degree-of-freedom are the longitudinal, lateral
and yaw motions [4],[2].
vxs = Tc−TrrIe f f
−Cxv2
xsm
+vys ψs
vys =Fy f
+Fyr−Cyv2ys
m−vxs ψs
ψs = 1Iz
[
l f Fy f− lrFyr
]
The characteristic of the tires is assumed to be linear
model where the lateral forces are proportional to the side
slip angle of the tire. The expression of these forces are
given as follows:{
Fy f= 2c f α f
Fyr = 2crαr
where{
α f = δ f−vys +l f ψs
vxs
αr = −vys−lrψs
vxs
The model can be rewritten in canonical form:
vxs = f0 + g0Tc (1)
vys = f1 + g1δ f (2)
ψs = f2 + g2δ f (3)
f0=−TrrIe f f
−Cxv2
xs
m+vys ψs
g0 = 1Ie f f
f1=−2 c f
m
vys + l f ψs
vxs− 2 cr
m
vys − lrψs
vxs−
Cyv2ys
m−vxs ψs
g1 =2c f
m
f2=−l f
Iz2c f
vys + l f ψs
vxs+ lr
Iz2cr
(
vys − lrψs
vxs
)
g2 = lrIz
2c f
(4)
In order to improve vehicle comportment, a kinematics
equations are formulated to express the longitudinal (inter
distance) and lateral displacement [3]. For more clearness,
we give some notations which will be useful thereafter. The
indices (l,s,r) will respectively represent the leader vehicle,
the follower vehicle and the relative variables.
dxr = −vxl+ dyrψl + vxs cosψr +
(
vys + ψsl f
)
sinψr
dyr = −vyl+(lr −dxr) ψl − vxs sinψr+
(vys + ψsl f )cosψr
(5)
The derivative of (5) give us:
{
dxr = a0 + b0Tc + c0δ f
dyr = a1 + b1Tc + c1δ f(6)
where
a0 =− vxl+ dyrψl + dyrψl + f0 cosψr − vxs sinψrψr
+(vys + l f ψs)cosψrψr +( f1 + f2l f )sin ψr
b0 =g0 cosψr
c0 =(g1 + g2l f )sin ψr
a1 =− dxrψl + dxrψl − (vyl− ψllr)− f0 sinψr
− (vys + l f ψs)sin ψrψr +( f1 + f2l f − vxsψr)cosψr
b1 =−g0 sinψr
c1 =(g1 + g2l f )cosψr
(7)
ψr = ψl −ψs (8)
III. SYNTHESIS AND OBSERVER DESIGN
In this section, we present the design of a sliding
mode observer scheme, based on algorithm of sub-optimal
control. This algorithm allows the finite time stabilization
of uncertain second order non linear with incomplete
state measurements [8],[10]. The main characteristic of
this observer is the robustness face to the strong non
linearities of the system and the parameters uncertainties
1141
of modeling. Our aim is to estimate the different unknown
vehicles states in order to make easier the automated
driving. First of all, let give in the following table the
different measurable and non measurable variables, where
the last column of table designate if the state is measurable
(m) or not measurable (nm).
vxl/vxs Longitudinal speed of vehicle leader/ follower m/m
ψl/ψs Yaw angle of vehicle leader/follower nm/m
ψl/ψs Yaw rate of vehicle leader/follower nm/m
dxr/dyr Longitudinal distance/ lateral displacement m/m
dxr/dyrVariation of longitudinal distance/
Variation of lateral displacementnm/nm
ψr Variation of relative heading nm
Table 1: Measurable and not measurable states
A. Estimation of variation relative yaw angle
It is assumed that the relative heading ψr for the two
vehicles is measured by a seteroscopic vision system. The
aim is to estimate the unknown state ψr from ψr. The
relative yaw dynamic can be expressed as follow:
{
x1 = x2
x2 = f2 + g2δ f(9)
with x1 = ψr and x2 = ψr. x2 is not measurable state, f2
and g2, are smooth functions are assumed uncertain. The
first step, we consider the following observer equations:
{ .ψr = vlacet r.vlacet r = ηr(t)
where ηr(t): R+ →R is an auxiliary input to be appropri-
ately selected to guarantee observer convergence in finite
time..ψr = vlacet r is the estimate of relative yaw angle varia-
tion. Now, the second step consists to define the observation
errors:
eψr = ψr − ψr (10a)
evlacet r= vlacet r − vlacet r (10b)
From the (10a) and (10b), we deduct the error dynamic
equations:{
eψr = evlacet r
evlacet r= ψr −ηr(t)
(11)
∣
∣ψl − f2 −g2δ f
∣
∣ ≤ M (12)
where M is a positive constant.
−ψsmax ≤ ψs = f2 + g2δ f ≤ ψsmax (13)
Now, the problem is to find ηr such that the errors (10a)
and (10b) are steered to zero in finite time.
Observer algorithm for the variation of the relative
yaw angle [10], [8]
i) Set γ∗ ∈ (0,1]
ii) Set eψr max = eψr(0)
Repeat, for t > 0, the following steps
iii) I f[
eψr(t)−12eψr max
][
eψr max − eψr(t)]
> 0 then
γ = γ∗ else set γ = 1
iv) I f eψr(t) is an extremal value then set eψr max = eψr(t)v) Apply the control law ηr(t) :
ηr(t) = −γUMaxsign(eψr(t)−1
2eψr max)
Until the end of the control time interval.
�
with UMax > max( Mγ∗ ; 4M
3−γ∗ )
The value of M is obtained empirically by using the
bounded states of vehicle characteristics
B. Estimation of yaw rate of vehicle leader
The yaw rate of the vehicle leader noticed ψl is not
measurable. As we know that the yaw rate of the vehicle
follower is measured by gyroscope. Therefor, we use the
estimate of the variation of the relative yaw angle, previ-
ously developed in order to deduce the variation of the yaw
angle for the leader.
.ψr =
.ψ l − ψs (14)
.ψ l =
.ψr + ψs (15)
C. Estimation of variation of the longitudinal distance
It is assumed that the longitudinal distance is obtained
by the LiDAR. In order to estimate the variation of the
longitudinal distance dxr, we should express the dynamic
of this variable.
Consider the following system{
x1 = x2
x2 = a0 + b0Tc + c0δ f(16)
where x1 = dxr et x2 = dxr.
The second equation of (18) is the same with equation (6),
where Tc et δ f are the control inputs and a0, b0, c0 smooth
functions.
The observer equations can writeen as follows:{ .
dxr = vxr.vxr = ηvxr (t)
.
dxr corresponds to the estimate of the variation longitu-
dinal distance, ηvxr (t) is auxiliary input which ensure the
convergence on finite time of the observer.
The observer errors are givern by:
edxr= dxr − dxr
evxr= dxr −
.
dxr
The equations of the error dynamic can be written:
{
edxr= evxr
evxr= a0 + b0Tc(t)+ c0δ f (t)−ηvxr
(t)
1142
The aim is to find out the positive constant M1
which satisfied this constraint:
∣
∣a0 + b0Tc + c0δ f
∣
∣ ≤ M1 (17)
In order to steer in finite time t the dxr to dxr. The bounded
of these functions a0,b0,c0 ensure the existence of M1.
The following table summarizes the various intervene in
the convergence of the algorithm.
γ∗ ∈ (0,1] γ∗ = 0.75
xMax = x1(0) edsr max = edxr
(0) = 10m
x1(t) edxr(t)
UMax UMax > max(M1γ∗ ;
4M13−γ∗ ) = 15
Table 2: Variables for estimation of the longitudinal
distance
D. Estimation of variation of the lateral displacement
It is assumed that the lateral displacement is given by the
video sensor. In this case we seek the lateral displacement.
The various steps can be summarized as:
Consider the following system{
x1 = x2
x2 = a1 + b1Tc + c1δ f(18)
with x1 = dyr and x2 = dyr.
The observer equation system can be written as:
{ .
dyr = vyr.vyr = ηvyr (t)
where.
dyr is the estimate of the variation of lateral displacement
and sηvyr(t) is the auxiliary input to be determined.
We consider the observer errors as:
edyr= dxr − dyr
evyr= dyr −
.
dyr
The dynamic error equations:
{
edyr= evyr
evyr= a1 + b1Tc + c1δ f −ηvyr
(t)
The input ηvyr(t) and the gain M2 are chosen such as:
∣
∣a0 + b0Tc + c0δ f
∣
∣ ≤ M2 (19)
The different numerical values of the algorithm is ad-
dressed in the following table:
γ∗ ∈ (0,1] γ∗ = 0.75
xMax = x1(0) edyr max = edyr
(0) = 0.05
x1(t) edyr(t)
UMax UMax > max(M2γ∗ ;
4M23−γ∗ ) = 5
Table 3: Variables for estimation of the lateral displacement
IV. SECOND ORDER SLIDING MODE CONTROL
In order to maintain safety distance, the lateral dis-
placement and relative yaw angle of the two vehicle are
controlled to trace their desired values by using second
order sliding mode. The aim is that the follower vehicle
should follow the motion of the leader vehicle.
A. Design methodology
Fig. 2. Inter distance model
In order to control the inter distance, we adopt a constant
headway time policy. The sliding surface is a function of
the inter distance model (see the figure 2):
Slong = dxr + dxrdes (20)
with dxr is the relative distance between the rear of the
leader vehicle and front of the follower vehicle.
dxrdes : reference distance is function longitudinal velocity
of vehicle follower.
dxrdes = d0 + hvxs , where h is the headway time 2 second1, d0 is stopping distance is equal to 2m. For the lateral
control, a vehicle following policy is chosen. The aim of
the control is to minimize a mixed criteria between follower
vehicle lateral displacement and relative yaw angle. This
criteria is: σlat = dyr + λ ψr.
λ is a weighting coefficient. The lateral surface is defined
with the following manner:
Slat = σlat + c1σlat + c2
∫ t
0σlatdτ (21)
The sliding surfaces are: S =[
Slong,Slat
]T. The equivalent
control inputs are computed as follow:
S =[
Slong, Slat
]T= 0.
S =
[
dxr + hvxs(
dyrλ ψr
)
+ a1
(
dyr + λ ψr
)
+ a2 (dyr + λ ψr)
]
(22)
S = G+ B
[
Tc
δ f
]
(23)
with
G =
[
dxr + h f0
a1 + λ ψl − f2 + c1
(
dyr + λ ψr
)
+ c2 (dyr + λ ψr)
]
and
B =
[
hg0 0
b1 c1 −λ g2
]
Since our aim is to apply the second order sliding mode.
1French law that specifies that drivers are not allowed to drive with aheadway time less than 2 seconds
1143
It offers good properties such as the elimination of the
chattering phenomenon. The Twisting Algorithm [11] is
applied to the derivative input δ f , Tc.
u =
−u |u| >∣
∣ueq
∣
∣
−KMsign(S) i f SS ≤ 0, |u| ≤∣
∣ueq
∣
∣
−kmsign(S) i f SS > 0, |u| ≤∣
∣ueq
∣
∣
(24)
Where KM and km are coefficients, determined to respect
the four conditions of applications of the twisting
Algorithm.
V. SIMULATION RESULTS
In this section, several simulations have been carried in
order to show the effectiveness of the proposed algorithm
for car-following problem.
A. Lance change maneuver
This maneuver is summarized as follows: the two ve-
hicles are located on the same right lane, are spaced of
a distance from 20 m evolve with the same longitudinal
speed of 16.66 m/s. The observer obviously synthesized is
used to estimate the non measurable variables which consist
of: the variation of relative yaw angle, variation of the
lateral displacement, variation of the longitudinal distance
and the yaw rate of the vehicle leader are illustrated on
figures (3,(b)), (3,(c)), (3,(d)), (3,(e)). We notice through
these figures, one simultaneously represented the measured
signal and its estimated. The following remarks can be
deduced:
• The estimated states converge quickly toward the real
states.
• The performances obtained are good as well in dy-
namics as in statics.
• The observation errors are steered to zero in finite time.
At time t = 0sec, the leader initiates a cut in maneuver
to the left lane. The aim of the follower vehicle is to make
a lane change maneuver by maintaining a safety distance
by acting on the steering angle of the front wheels and the
composite torque. However, it has also to perform braking
as the actual spacing is smaller by 15.32m than the desired
spacing of 35.32m which corresponds to a headway time
of 2sec. The profile of lane change maneuver is carried
out successfully with small steady state error (see figure
(4,(a)).The evolution in time of the sliding manifold which
correspond to the spacing error (4d). It can be noticed the
convergence towards zero which proves the reached desired
spacing.
B. Stop and Go maneuver
We want, through this maneuver, to give an outline of
the daily traffic car following in a case of stop and go.
Initially, the follower vehicle is located on the left lane
and the vehicle leader is located on the right lane, while
the inter distance is 27.5m. The vehicle leader moves
with an initial velocity of 36km/h, the following vehicle
evolves with initial velocity of 60km/h. Since the desired
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5-1
-0.5
0
0.5
1
1.5
2
Derivative of Slat
Sla
t
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5-1
-0.5
0
0.5
1
1.5
2
Derivative of Slat
Sla
t
0 2 4 6 8 10 12 14 16 18 20-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
Time (sec)
Variation o
f re
lative y
aw
angle
(R
ad/s
)
Estimated
Measured
(a) (b)
0 2 4 6 8 10 12 14 16 18 20-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
Time (sec)
yaw
rate
of
vehic
le leader
(Rad/s
)
0 2 4 6 8 10 12 14 16 18 20-6
-5
-4
-3
-2
-1
0
1
Time (sec)
Variation o
f lo
ngitudin
al dis
tance (
m/s
)
Estimated
Measured
(c) (d)
0 2 4 6 8 10 12 14 16 18 20-1
-0.5
0
0.5
1
1.5
Time (sec)
Variation o
f la
tera
l dis
pla
cem
ent
(m/s
)
Estimated
Measured
(e)
Fig. 3. Profile of lane change maneuver. (a) : Sliding manifold (Slat ,Slat),(b): Variation of relative yaw angle, (c): Yaw rate of vehicle leader, (d):Variation of longitudinal distance, (e): Variation of lateral displacement
0 50 100 150 200 250 300 350-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Longitudinal position (m)
Latr
al positio
n (
m)
0 2 4 6 8 10 12 14 16 18 20-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Time (sec)
Ste
ering a
ngle
(deg)
0 2 4 6 8 10 12 14 16 18 20-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Time (sec)
Ste
ering a
ngle
(deg)
(a) (b)
0 2 4 6 8 10 12 14 16 18 20-1.5
-1
-0.5
0
0.5
1
1.5
Time (sec)
Longitudin
al accele
ration (
m/s
2)
0 2 4 6 8 10 12 14 16 18 20-1.5
-1
-0.5
0
0.5
1
1.5
Time (sec)
Longitudin
al accele
ration (
m/s
2)
0 2 4 6 8 10 12 14 16 18 20-35
-30
-25
-20
-15
-10
-5
0
5
Time (sec)
Spacin
g e
rror
(m)
0 2 4 6 8 10 12 14 16 18 20-35
-30
-25
-20
-15
-10
-5
0
5
Time (sec)
Spacin
g e
rror
(m)
(c) (d)
Fig. 4. Profile of lane change maneuver: (a) : lateral and longitudinalpositions of the leader (*,-.) and the follower (o,-), (b): Steering angle,(c): Longitudinal acceleration of vehicle follower, (d): Spacing error
1144
longitudinal distance is of 35.32m. In order to control this
distance, the follower must decreases while decelerating
to reach the speed of 35km/h and at the same time, the
vehicle leader accelerates to reach at the end of 10 seconds
the value of 60km/h and makes a change of heading
direction at the distance of 60 m with a yaw angle ψl= 0.1
rad, then it continues its trajectory with this same speed.
To preserve the desired inter distance, the follower vehicle
accelerates one again to reach the speed of 60km/h. At t=
14sec, the vehicle leader carries out a deceleration until
to stop, and thereafter at t = 40sec the vehicle accelerates
to reach a velocity of 60km/h with this same speed for
one duration of 10sec. The leader decelerates (60km/h to
35km/h) and finish the trajectory by an uniform movement
of 36km/h. The velocities profile are given on figure (5,a).
We can note that this maneuver is carried out with
success with respect to the passengers comfort where the
reasonable boundaries in acceleration range are from -0.3g
to 0.2g as depicted on figure (5,d). It can be seen that the
desired spacing distance is reached (see figure (5,c)).
According to the figures (5,(e),5,(f)), which correspond
to the variation of the lateral displacement, the variation
of longitudinal distance, we can also note that both plots
are great identical which proves the convergence of the
observer in a finite time with a negligible steady state
error and the estimated curve follows the measured one.
0 10 20 30 40 50 60 70 800
10
20
30
40
50
60
Time (sec)
Longitudin
al speed (
Km
/h)
0 10 20 30 40 50 60 70 80-1500
-1000
-500
0
500
1000
1500
Time (sec)
Com
posite torq
ue (
N.m
)
0 10 20 30 40 50 60 70 80-1500
-1000
-500
0
500
1000
1500
Time (sec)
Com
posite torq
ue (
N.m
)
(a) (b)
0 10 20 30 40 50 60 70 80-10
-8
-6
-4
-2
0
2
Time (sec)
Spacin
g e
rror
(m)
0 10 20 30 40 50 60 70 80-4
-3
-2
-1
0
1
2
3
Time (sec)
Longitudin
al A
ccele
ration (
m/s
ec
2)
0 10 20 30 40 50 60 70 80-4
-3
-2
-1
0
1
2
3
Time (sec)
Longitudin
al A
ccele
ration (
m/s
ec
2)
(c) (d)
0 10 20 30 40 50 60 70 80-2
-1.5
-1
-0.5
0
0.5
Time (sec)
Variation o
f la
tera
l dis
pla
cem
ent(
m/s
)
Estimated
Measured
0 10 20 30 40 50 60 70 80-6
-4
-2
0
2
4
6
8
10
12
Time (sec)
Variation longitudin
al dis
tance(m
/s)
Estimated
Measured
(e) (f)
Fig. 5. Stop and Go maneuver. (a): Leader (-.) and follower (-) longitu-dinal speed, (b): Composite torque, (c): Spacing error, (d): Longitudinalacceleration of follower vehicle, (e): variation of lateral displacement,(f):variation of longitudinal distance
VI. CONCLUSION
A non linear control approach for the vehicle following
problem has been developed. By estimating the non mea-
surable states with sliding mode observer. This observer
is able to build the non measurable states. Afterwards,
these different estimated states were implemented on the
integrated longitudinal and lateral control problem. The
observer and control law are based on a variable structure
theory. Simulations have been carried out to illustrate
the ability of this approach to give well performance of
the states estimation and control law synthesis in some
scenarios of automated driving.
REFERENCES
[1] M. Mangeas, A priori assessment of driver assistance system with
respect to capacity and safety, Intelligent Vehicle Symposium IV2001, Tokyo, June 2001.
[2] A. Chaibet, L. Nouveliere, S. Mammar, M. Netto, Backstepping
Control Synthesis For Both Longitudinal And Lateral Automated Low
Speed Vehicle, IEEE, Intelligent Vehicles, IV 05, Las vegas, 2005.[3] A. Chaibet, Controle Lateral et Longitudinal pour le Suivi
de Vehicule, PhD thesis, IBISC-LSC, Universite D’Evry ValD’Essonne, 2006.
[4] D. Swaroop, Seok Min Yoon, The design of a controller for vehicle
following in an emergency lane change maneuver, California PATHWorking Paper UCB-ITS-PWP-99-3, 1999.
[5] Y. Sebsadji,S. Glaser, S. Mammar, J. Dakhlallah, Road Slope andVehicle Dynamics Estimation, American Control Conference, 2008.
[6] H. Cherouat, S. Diop, An observer and an integrated braking/traction
and steering control for a cornering vehicle, American ControlConference, 2005.
[7] S. Mammar, D. Koenig, Vehicle Handling improvement by Active
Steering, Vehicle System Dynamics, vol 38, No3, 211-242, 2002.[8] G. Bartolini, A. Ferrara E. Usai, Applications of a Sub Optimal Dis-
continuous Control Algorithm for Uncertain Second Order Systems,International Journal of Robust and Nonlinear Control, 7, 4, 297-310,1997.
[9] J. Bom, B. Thuilot, F. Marmoiton and P. Martinet, A GlobalControl Strategy for Urban Vehicles Platooning relying on Nonlinear
Decoupling Laws, IEEE/RSJ-IROS’05, Alberta, 2005.[10] A. Ferrara P. Pisu, Minimum Sensor Second Order Sliding Mode
Longitudinal Control of Passenger Vehicles, IEEE Transactions onIntelligent Transportation Systems, 5,1, March 2004.
[11] L. Nouveliere, S. Mammar, Experimental Vehicle Longitudinal Con-
trol using Second Order Sliding Modes, American Control Confer-ence, 2003.
[12] L. Nouveliere, Commandes robustes appliquees au controle assiste
d’un vehicule a basse vitesse, These pour l’obtention du doctorat del’Universite de Versailles, Dec. 2002.
1145