Vous êtes sur la page 1sur 10

International Journal of Electrical and

Electronics Engineering Research (IJEEER)


ISSN(P): 2250-155X; ISSN(E): 2278-943X
Vol. 6, Issue 4, Aug 2016, 59-68
TJPRC Pvt. Ltd.

LINEAR CONTROL OF VEHICLES LATERAL DYNAMICS


DER-CHERNG LIAW, WEN-CHING CHUNG & CHIEH TSAO
Department of Electrical Engineering, National Chiao Tung University, Hsinchu, Taiwan, R.O.C
ABSTRACT
In this paper, control issue for vehicle dynamics is studied. Controllability for the vehicle dynamics is first
discussed. A state feedback control scheme is then applied to prevent and/or delay the appearance of saddle-node
bifurcation. Case studies demonstrate the effectiveness of proposed design. Two-Parameter bifurcation analysis of
closed-loop system is also obtained to classify the regime of the effective control gains for system stabilization.
KEYWORDS: Stabilization, Vehicle, Control, Bifurcation

Received: May 02, 2016; Accepted: May 18, 2016; Published: Jul 21, 2016; Paper Id.: IJEEERAUG20167

1. INTRODUCTION
In the recent years, the study of the dynamical behavior of vehicle dynamics have attracted lots of

possible causes of making traffic accidents and enhance the driving safety for automated steering vehicle. It is
known that traffic accidents might be highly related to the nonlinear behavior of vehicle dynamics. For instance,
the vehicle is easier going into spin when a steering wheel is carelessly operated. The study of the linkage between
the nonlinear phenomena of vehicle dynamics and the applied front wheel steering angle hence becomes a very
important issue. Among those existing studies (Ackerman, et al (1995)) adopted the sliding mode approach to

Original Article

attention (e.g., Wong (2001), Ackerman, et al (1995), Ray (1995), Ono (1998), Chung and Yi (2006)) to find

design robust control laws for providing system stability with respect to the large variation of system parameters
such as axial velocity, mass of the vehicle and the contact force between tire and road surface. Based on a five
degree of freedom vehicle model, an extended Kalman filter (EKF) was proposed to estimate the historic data of
vehicle motion and tire force (Ray (1995)). A linearized state-space model of the lateral vehicle dynamics was
adopted to design robust automatic steering control for look-down reference systems with front and rear sensors
(Guldner et al (1999)). The linkage between saddle-node bifurcation and system instability in vehicle dynamics
was observed in (Ono (1998)) via a numerical example. It is known that the appearance of saddle node bifurcation
in a nonlinear system implies that the system might lose its continuous equilibrium branch as system parameters
vary (Guckenheimer and Holmes (1983), Chow and Hale (1992)). Such a bifurcation phenomenon might result in
the spin of a vehicle as steering angle changes. The application of bifurcation theory to engineering problems has
been well exploited in the recent years (e.g., Liaw et al (2007), Liaw and Abed (1996), Liaw and Song (2001)).
A theoretical analysis of vehicle dynamics has been obtained by applying the bifurcation theory (Liaw et al
(2007)), which revealed that the uncontrolled version of vehicles lateral dynamics might undergo a stationary
saddle-node bifurcation. In addition, the controllability and direct state feedback control laws were also proposed
in (Liaw and Chung (2006)) to stabilize the bifurcation points. In this paper, we continue the study of our previous
work as those in (Liaw et al (2007)) and (Liaw and Chung (2006)) but emphasizing the effects of linear control
gain on the stabilization design of vehicles lateral dynamics at the bifurcation point. It is achieved by the
www.tjprc.org

editor@tjprc.org

60

Der-Cherng
Cherng Liaw, Wen-Ching
Wen
Chung & Chieh Tsao

numerical study of parametric bifurcation analysis of the overall vehicle dynamics after adding
addin control.
The paper is organized as follows. Mathematical models of vehicle dynamics are recalled in Section 2.
It is followed by the feedback stabilization design at the saddle-node
saddle node bifurcation. Finally, simulation results for an example
exa
vehicle model are given in Section 4 to demonstrate the analytical results. Parametric bifurcation analysis of vehicle
dynamics is also carried out to exploit possible effects of controller gain on systems nonlocal behavior.

2. VEHICLE DYNAMICS
In the
he following, we will recall the mathematical model for vehicles steering dynamics from(Wong
from(
(2001)).
Those results will then be adopted in Section 3 for the two parameter analysis of system performance in the selection of the
control gains.
Consider the vehicles steering dynamics as depicted in Figure 1 (e.g., Wong (2001),
(2001) Liaw et al (2007)), where

L f (resp. Lr ) denotes the length between the center of gravity (CG) and front-wheel
front wheel axes (resp. rear-wheel
rear
axes).
In addition,

Fyfl and Fyrl (resp. Fyfr and Fyrr ) are the left-side (resp. right-side)
side) cornering forces of front and rear tires.

Moreover, both Fxfl and

Fxrl (resp. Fxfr and Fxrr ) are the left-side


side (resp. right side) traction forces of front and rear tires.

Note that, all of those cornering forces are functions of sideslip angle and yaw rate.

Figure 1: Vehicles Steering Dynamics.


By assuming the vehicle body is symmetric about the longitudinal plane, the basic motion equations of steering
dynamics without considering roll motion were derived in (Wong
(
(2001)) as given by

m(& ) = Fxf + Fxr Fyf sin f ,

(1a)

m ( & + ) = Fyf + Fyr + Fxf sin f ,

(1b)

I z& = ( L f Fyf Lr Fyr ) cos + L f Fxf sin f ,

(1c)

where,

and

denote sideslip angle and yaw rate as system outputs, respectively, and

angle as only system input. In addition, m,

Impact Factor (JCC): 6.1843

is the front-wheel

I z and are the mass of the vehicle, yaw moment around z-axis
z
and

NAAS Rating: 2.40

Linear Control of Vehicles Lateral Dynamics

longitudinal velocity, respectively, while Fyf =

61

Fyfl + Fyfr , Fyr = Fyrl + Fyrr , Fxf = Fxfl + Fxfr and Fxr = Fxrl + Fxrr .

In this study, we focus on the characteristic analysis of lateral dynamics by assuming

Fyf = 0 and & = 0. Note that, one possible application of such a concern is the steering control of vehicle under
cruise control mode. Thus, the steering dynamics as in Eq. (1) with constant speed

can then be reduced to a second-

order model as given by

& = h1 ( , ) =
& = h2 ( , ) =
Let

1
m

( Fyf + Fyr ) ,

(2a)

1
( L f Fyf Lr Fyr ) cos .
Iz

(2b)

xe = ( e , e )T be an equilibrium point of system (2) for a given steering wheel angle f = f ,e .

The condition of equilibrium point can then be solved as

Fyf ( xe , f ,e ) + Fyr ( xe , f ,e ) = m e

(3)

and

L f Fyf ( xe , f ,e ) Lr Fyr ( xe , f ,e ) = 0

(4)

or cos e = 0

(5)

It is clear from Eq. (5) that the solution of

can be easily obtained as n +

vehicle cannot easily achieve such a condition. Thus, the equilibrium point

for integer n. However, a

xe should satisfy both conditions of Equations

(3) and (4). By using Routh Hurwitz stability criterion, stability condition of the equilibrium point

xe = ( e , e )T was

obtained in (Liaw et al (2007)) as recalled below.


Lemma 1 (Liaw et al (2007)) The equilibrium point
and

xe of system (2) is asymptotically stable if a11 + a22 < 0

a11 a22 a12 a21 > 0, where ai s are given in Eqs. (6)-(9) below:

(6)

(7)

(8)

www.tjprc.org

editor@tjprc.org

62

Der-Cherng Liaw, Wen-Ching Chung & Chieh Tsao

(9)
The existence conditions of the saddle node bifurcation for system (2) have also been obtained in (Liaw et al
(2007)) and are recalled as follows.
Lemma 2 (Liaw et al (2007)) The system (2) undergoes the saddle node bifurcation at the equilibrium point

xe if

the following three conditions hold:

a11a22 = a12 a21 with a11 + a22 < 0,

(10a)

a22b1 a12b2 , and

(10b)

a22 q11 a21q12 +

2
a21
a2
q13 a12 q21 + a11q22 11 q23 0
a22
a12
,

(10c)

where

b1 =

F
1 Fyf
[
( xe , f ,e ) + yr ( xe , f ,e )],
f
mv f

(10d)

b2 =

F
F
cos e
[ L f yf ( xe , f ,e ) + Lr yr ( xe , f ,e )],
Iz
f
f

(10e)

with

q11 =

q22 =

2 h1
2h1
2 h2
2h1
q
=
(
x
,

)
,
x
q
=
(
,

),
q
=
(
x
,

),
( x , ),
2 e f ,e 12 e f ,e 13 2 e f ,e 21 2 e f ,e

2h
2 h2
( xe , f ,e ) ,and q23 = 2 2 ( xe , f ,e ) .

3. MAIN RESULTS
In this section, control issue for vehicle dynamics is studied. Result of system controllability for the vehicle
dynamics is first studied. State feedback control scheme is studied to prevent and/or delay the appearance of saddle-node
bifurcation. It is followed by case study to demonstrate the effectiveness of proposed design. Two-Parameter bifurcation
analysis of closed-loop system is also obtained to classify the regime of the effective control gains for system stabilization.
As discussed in Lemma 2, system (2) could undergo stationary saddle-node bifurcation, which might result in spin
of vehicle. In order to prevent the appearance of such an instability, in the following we will seek for possible control laws
for system stabilization. First, we discuss the controllability of system (2).
By letting

x = ( , )T , x% = x xe and the control input u% = f f ,e , we then have the linearization of

system (2) at xe and f

= f ,e as

Impact Factor (JCC): 6.1843

NAAS Rating: 2.40

Linear Control of Vehicles Lateral Dynamics

63

x%& = Ax% + Bu% ,

(11)

where

a12
b1
and B = .

a22
b2

a
A = 11
a21

Here, the values of all ai j and bi for i,j=1 and 2 are defined above.
The controllability matrix C of the linearized model can then be obtained as

b
C= 1
b2

a11b1 + a12b2
.
a21b1 + a22b2

(12)

The determinant of the controllability matrix C is calculated as

det(C ) = a21b12 a12b22 + ( a22 a11 )b1b2

(13)

The next result is recalled from (Liaw and Chung (2006)).


Lemma

The

linearized

model

(11)

is

controllable

at

the

equilibrium

point

xe

if

a21b12 a12 b22 + ( a22 a11 )b1b2 0 .


Remark 1 we consider a special case of which
have

xe being the saddle-node bifurcation point. From Lemma 2, we

a11a22 a12 a21 = 0 and a22b1 a12b2 when the linearized model is evaluated at the saddle-node bifurcation

point. Eq. (12) can then be rewritten as

det[C ] =

1
(a22b1 a12b2 )(a11b1 + a12b2 ) .
a12

(14)

The next result follows readily from Lemma 3.


Corollary 1 The system (2) is controllable at the stationary saddle-node bifurcation point

xe if

a11b1 + a12b2 0 .
As discussed in Corollary 1, the saddle-node bifurcation point of system (2) is controllable if

a11b1 + a12b2 0 .

This leads to the possibility of a design of state feedback control law for system stabilization. Let the control input

u% = K d x% , where K d = [kd 1 , kd 2 ] . The linearized model (11) with state feedback controller can then rewritten
as

x&% = ( A + BK d ) x% .

(15)

The characteristic equation of system (15) is calculated as

www.tjprc.org

editor@tjprc.org

64

Der-Cherng Liaw, Wen-Ching Chung & Chieh Tsao

2 + 1 + 2 = 0 ,

(16)

where

1 = kd 1b1 + kd 2b2 a11 a22 ,

(17a)

2 = a11a22 a12 a21 + (a12b2 a22b1 )kd 1 + (a21b1 a11b2 )kd 2 .

(17b)

By applying the Routh-Hurwitz stability criterion to the characteristic equation of system (15), we then have the
following stabilization result for equilibrium point xe .
Theorem 1 The equilibrium point

xe of system (2) is asymptotically stabilizable if 1 > 0 and 2 > 0, where

1 and 2 are given in Eq. (17).


xe being the saddle-node bifurcation point, we have

Remark 2 Consider a special case of which

a11a22 = a12 a21 and a22b1 a12b2 , which implies a21b1 a11b2 . The value of 2 can then be reduced to
(a12b2 a22b1 )kd 1 + (a21b1 a11b2 )kd 2 . Since a22b1 a12b2 and a21b1 a11b2 , it will not be difficult to find kd 1 and
kd 2 for 2 > 0 . In addition, it is observed from Eqs. (10d) and (10e) that at least one of b1 and b2 will be nonzero.
Thus, we can have

1 > 0

by proper choice of

kd 1 and kd 2 , which implies that the stabilizing controller for system (2) at

the saddle-node bifurcation point does exist. The next result follows readily from Theorem 1.
Corollary 2 The stationary saddle-node bifurcation point

xe of system (2) is asymptotically stabilizable.

As presented in (Liaw and Chung (2006)), the saddle-node bifurcation points will be a function of vehicle speed.
To demonstrate the effectiveness of the proposed designs above, in this paper we will construct linear control laws for
*

system stabilization at the bifurcation points xe =

( e , e )T = (0.012, 0.2275)T with f = 0.0569 for vehicle

speed v = 10 m/s.
*

First, the determinant of the controllability matrix C for vehicle system at the bifurcation point xe is calculated as

det[C ] = 209.211 0 . This implies that the equilibrium point xe* is controllable. Next result follows readily from
Theorem 1.
*

Corollary 3 The saddle-node bifurcation point xe of system (2) is asymptotically stabilizable by the linear
control law if -2.5276-5.9996 kd 2 <

kd1 < 0.0056+2.9267 kd 2

with

kd 2 > 0.2838 .

Figure 1 shows the time response for the uncontrolled system at

v = 10 m/s with initials 0 = 0.01 and 0 = -0.24,

which is near the equilibrium point SNBP01. It is observed from Figure 1 that the values of

and

about 5 seconds, which is very dangerous in practical driving. According to Corollary 3, we choose

Impact Factor (JCC): 6.1843

become large after

kd 2 = 0.1 or 1. It is

NAAS Rating: 2.40

Linear Control of Vehicles Lateral Dynamics

65

clear from Corollary 3 that the saddle-node bifurcation point xe of system (2) is asymptotically stable for -3.1276 < kd 1
*

0.2982 at

kd 2 = 0.1 or -8.5272 < kd1 < 2.9323 at kd 2 = 1. Figures 2 and 3, respectively, show the time responses of the

closed-loop system being stable and unstable for different choice of


of

<

kd 1.

It is observed form Figure. 2 that the oscillations

and are substantially improved as kd1 decreases.

Define the total energy of control force TEu% = u% dt , and the maximum amplitude of control force

MAu% = max | u% | , respectively. The values of TEu%


1. For

and

MAu%

for different control gains at

v =10 m/s are given in Table

kd 2 = 0.1, the values of TEu% and MAu% are decreasing as kd1 increases from -2.5 to -0.5. However, the values of

TEu% and MAu% for kd 2 = 1 are increasing as kd1 varies from -2.5 to 0.5.
Table 1 Energy Indices for

kd 2
0.1

v = 10 m/s

kd1

TEu% ( 105 )

MAu% ( 102 )

-2.5
-1.5
-1
-0.5
-5
-2.5
-0.5
0.5

1.3923
0.2523
0.1228
0.0590
0.3985
0.3318
0.3994
0.4488

0.5088
0.2459
0.1585
0.0907
0.4601
0.7382
1.1476
1.3524
*

The bifurcation analysis near the saddle-node bifurcation point xe is obtained as shown in Figure.4. Note that, in
Figure 4 the symbol uc denotes the bifurcated solutions for the uncontrolled system, square-box denotes the saddle-node
bifurcation point and triangle box denotes the Hopf bifurcation point, respectively. Figure 4 reveals that the saddle-node
bifurcation point of vehicle system is successfully postponed for
-2.5 and -5 with

kd1 = -1, -1.5 and -2.5 with kd 2 = 0.1, and for kd1 = -0.5,

kd 2 = 1.

Two-parameter bifurcation diagrams of the closed-loop system with

kd 2 = 0.1, 0.5 or 1 are also obtained by using

code AUTO and depicted in Figure 5, which indicates the location of saddle node bifurcation points (dotted-line) and Hopf
bifurcation points (solid line). The dashed-line denotes the separated line. Figures 5 (d)-(f) are the zoom in of Figures 5(a)(c), respectively. It is observed from Figure 5 that the number of saddle node bifurcation will be different as
There are two more saddle-node bifurcation points found for -1.0172 kd 1 < 32.8658 at
67.4020 at

kd1 varies.

kd 2 = 0.1, -1.0206 kd1 <

kd 2 = 0.5 and -1.0228 kd1 < 101.7270 at kd 2 = 1. Moreover, there are four saddle-node bifurcation points

found for -1.2605

kd1 -1.0172

at

kd 2 = 0.1, -1.4915 kd1 -1.0206 at kd 2 = 0.5 and -1.7816 kd1 -1.0228 at

kd 2 = 1. Moreover, it is observed from Figure 5 that the Hopf bifurcation point is the lower bound of kd1 for stable control
design. As shown in Figure 5, two-parameter bifurcation diagram can be divided into different operation zone. Table 2
summarizes the number of equilibrium points and their stability in each zone.
www.tjprc.org

editor@tjprc.org

66

Der-Cherng Liaw, Wen-Ching Chung & Chieh Tsao

Table 2 Number of Equilibrium Point in Figure 5


Zone
a
b
c
d

Stable Equilibrium Point


0
1
1
2

Unstable Equilibrium Point


1
0
2
1

4. CONCLUSIONS
In this paper, we have further studied the stabilization results presented in (Liaw and Chung (2006)) to more focus
on the post-control dynamical behavior near the saddle-node bifurcation point of the uncontrolled vehicle system.
Two-parameter bifurcation analysis is obtained to reveal the effects of linear control gain on post-control system stability.
Those results might provide a guide in the practical application.
REFERENCES
1.

Wong, J. Y. (2001). Theory of Ground Vehicles. John Wiley & Sons, Inc.

2.

Ackerman, J., Guldner, J., Sienel, W., Steinhauser, R., Utkin, V.I. (1995). Linear and nonlinear controller design for robust
automatic steering, IEEE Trans. Control Systems Technology, 132-143.

3.

Ray, L.R. (1995). Nonlinear state and tire force estimation for advanced vehicle control, IEEE Trans. Control Systems
Technology, 117-124.

4.

Ono, E., Hosoe, S., Tuan, H.D., Doi, S. (1998). Bifurcation in vehicle dynamics and robust front wheel steering control, IEEE
Trans. Control Systems Technology, 412-420.

5.

Guldner, J., Sienel, W., Tan, H.-S., Ackermann, J., Patwardhan, S., B{u}nte, T. (1999). Robust automatic steering control for
look-down reference systems with front and rear sensors, IEEE Trans. Control Systems Technology, 7(1), 2-11.

6.

Chung, T., Yi, K. (2006). Design and evaluation of side slip angle-based vehicle stability control scheme on a virtual test track,
IEEE Trans. Control System Technology, 14(2), 224-234.

7.

Guckenheimer, J., Holmes, P. (1983). Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields,
Springer-Verlag, New York.

8.

Chow, S. N., Hale, J. K. (1992). Methods of Bifurcation Theory, Springer-Verlag, New York.

9.

Liaw, D.-C., Chiang, H.-H., Lee, T.-T. (2007). Elucidating vehicle lateral dynamics using a bifurcation analysis. IEEE Trans.
Intelligent Transportation Systems.

10. Liaw, D.-C., Abed, E. H. (1996). Active control of compressor stall inception: a bifurcation-theoretic approach, Automatica,
32(1), 109-115.
11. Liaw, D.-C., Song, C.-C. (2001). Analysis of longitudinal flight dynamics: a bifurcation-theoretic approach. J. Guidance,
Control, and Dynamics, 24(1), 109-116.
12. Liaw, D.-C., Chung, W.-C. (2006). Control design for vehicle's lateral dynamics, 2006 IEEE International Conference on
Systems, Man, and Cybernetics, Taipei, Taiwan, R.O.C., 8-11 Oct.

Impact Factor (JCC): 6.1843

NAAS Rating: 2.40

Linear Control off Vehicles Lateral Dynamics

67

APPENDICES

Figure 1:
1 Time Responses for the Uncontrolled System

Figure 2: Time Responses for State Feedback Control System with Stable Equilibrium Point

Figure 3: Time Responses for State Feedback Control System with Unstable Equilibrium Point
www.tjprc.org

editor@tjprc.org

68

Der-Cherng Liaw, Wen-Ching Chung & Chieh Tsao

Figure 4: Bifurcation Diagram for State Feedback Control System Near xe

Figure 5: Two-Parameter Bifurcation Diagram of

Impact Factor (JCC): 6.1843

kd1 v.s. f

NAAS Rating: 2.40