Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

SAE TECHNICAL
PAPER SERIES 2003-01-1321

Parametric Study on Vehicle-Trailer Dynamics

for Stability Control
Weiwen (Kevin) Deng
General Motors Corporation

Xiaodi Kang
Mokum Tech. Inc.

Reprinted From: Vehicle Dynamics & Simulation, 2003

(SP-1778)

2003 SAE World Congress

Detroit, Michigan
March 3-6, 2003

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760 Web: www.sae.org
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.

For permission and licensing requests contact:

SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: [email protected]
Fax: 724-772-4028
Tel: 724-772-4891

For multiple print copies contact:

SAE Customer Service

Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-1615
Email: [email protected]

ISSN 0148-7191
Copyright © 2003 SAE International

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.

Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE.

Printed in USA
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

2003-01-1321

Parametric Study on Vehicle-Trailer

Dynamics for Stability Control
Weiwen (Kevin) Deng
General Motors Corporation

Xiaodi Kang
Mokum Tech. Inc.

Copyright © 2003 SAE International

ABSTRACT Most work done so far has been based on simplified

linear models with small hitch (articulation) angle
For stability control on a vehicle-trailer combination, the assumption to facilitate the linear controller design and
coupling between vehicle and trailer as two pivoted rigid analysis without appropriate justification. This paper
bodies has been investigated, in view of concerns that further confirms the validity of the linear model of a
the coupling strength varies as vehicle-trailer parameters vehicle-trailer combination through linearization to a
change, and thus plays an important role in the design simplified nonlinear model at various hitch angle
and robustness of the stability controller. Analyses equilibrium points and analysis on its characteristics as
through root locus and time-domain response have been affected by the various operating points.
conducted and several observations have been made,
which are valuable for the controller design. A Based on parametric studies, a preliminary control
preliminary stability controller with state feedback has method using state feedback is adopted for vehicle-
been designed and simulated to further verify the studies trailer system stability control. Computer simulation
above. based on a nonlinear full vehicle-trailer model is further
performed to confirm the results from the linearized
INTRODUCTION analysis.

Directional stability has been the primary concern of MODEL DESCRIPTION

vehicle-trailer combinations, especially at high speed.
Understanding the system characteristics of vehicle- MODELING ASSUMPTIONS
trailer combinations is the foremost step towards a good
approach of stability control. While studies on vehicle- x The vehicle and trailer bodies are symmetric about
trailer combinations have been widely conducted [1-4], their respective longitudinal planes.
this paper is to address two issues which were little x Only yaw plane motion with constant forward speed
explicitly discussed in the literature: mechanical coupling is considered, roll, pitch and bounce motions are
between vehicle and trailer, and hitch angle effect at neglected.
various operating points. x External lateral forces acting on the vehicle-trailer
system are limited to tire cornering forces, which can
The approaches on stability control of vehicle-trailer be expressed as linear functions of the
combinations proposed so far have been active steering corresponding tire sideslip angles and cornering
and/or braking from vehicle and/or trailer [5-7]. The stiffness. Tire aligning torques are small enough to
effectiveness of the control variable (input) from vehicle be omitted. Hitch pivot friction is negligible.
to trailer, or vise versa, is greatly determined by the x Small steering angles and sideslip angles. All
strength of the mechanical coupling between the two products of variable terms (yaw rate and sideslip
units. A weakly coupled system imposes weak influence velocity) are small and thus are ignored.
from one to the other. Such coupling strength is mainly
related to the system geometric and dynamic Figure 1 illustrates the bicycle model of an all-wheel
parameters. This paper discusses how such coupling or steering vehicle-trailer system. The original model
interaction is influenced by the parameter variation considered here was based on Jindra's model [1]. The
based on which the control strategy and parameters degrees of freedom (DOF) considered include vehicle
need to be adjusted accordingly. lateral motion, and vehicle and trailer yaw motions.

1
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

Y
vf
Df x
y

Fyf
v u
Gf
vr o
a1
Dr r

Fyr b1
x Gr
c
vt
T
Dt a2

Fyt b2

Gt

O X

Figure 1: An all wheel steering vehicle-trailer system model with three degrees of freedom

EQUATIONS OF MOTION The variables u, v, r, and T denote the vehicle forward

speed, the vehicle sideslip velocity, the vehicle yaw rate
With the sign conventions defined in the Appendix, the and the hitch angle, respectively; vf, vr and vt are the
lateral and yaw equations of motion for the vehicle in the speeds at vehicle front and rear tires as well as trailer
vehicle-body coordinate system can be directly obtained tire, respectively; Gf, Gr and Gt represent the steering
from the summation of lateral forces and yaw moments angles of vehicle front, vehicle rear and trailer wheels,
about the vehicle center of gravity (CG) as: respectively. For a four-wheel-steering (4WS) vehicle-
trailer system, the trailer steering angle Gt=0.
m1 (v  ur ) Fyf  Fyr  Fyh
(1)
I z1 r Fyf a1  Fyr b1  Fyh c Elimination of the coupling reaction forces and
substitution of the cornering forces as functions of
where Fyf and Fyr are the total cornering forces at the
sideslip angles yield three dynamic equations of motion
vehicle front and rear wheels, respectively, while Fyh is
of the vehicle-trailer system, which is presented in the
the lateral component of the coupling force between the
Appendix. Comparison of the vehicle-trailer model
vehicle and the trailer.
simulation results with those from testing further
demonstrates the validity of the simplified model and
Similarly, the equations of motion for the trailer in the
selected parameters.
vehicle-body coordinate system are expressed as:
EIGENVALUE ANALYSIS
0 Fxh  Fyt sin(T  G t )
m2 [v  ur  cr  a2 (r  T) cosT ] Fyt cos(T  G t )  Fyh (2) To design a suitable and effective stability controller, it is
essential that the vehicle-trailer system dynamic
I z 2 (r  T) Fyt b2  b2 (Fyh cosT  Fxh sinT ) characteristics under various operating conditions be
explored.
where Fxh is the longitudinal component of the coupling
force between the vehicle and the trailer. The other As evident from the equations of motion of the simplified
parameters in (1) and (2), m1, m2 and Iz1, Iz2, are, vehicle-trailer system listed in the Appendix, the hitch
respectively, the masses and yaw moments of inertia for angle is the only variable that induces non-linearity in the
the vehicle and the trailer; a1 and b1 are the distances system. For small hitch angle operations, such as small
from the vehicle CG to the front axle and rear axle, perturbations during a straight-line motion, the vehicle-
respectively; c denotes the distance from the vehicle CG trailer combination is a linear system. In the following
to the hitch point; a2 and b2 are the distances from the subsections, the eigenvalue characteristics of the
trailer CG to hitch point and trailer axle, respectively. vehicle-trailer system will be evaluated under both near
2
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

straight-line motion and various different steering Figure 2: Vehicle and vehicle-trailer system root loci as functions
maneuvers. of forward speed (umax=120 km/h)

7
VEHICLE-TRAILER SYSTEM STABILITY Vehicle only
CHARACTERISTICS DURING STRAIGHT-LINE Vehicle with trailer (1)
Vehicle with trailer (2)
MOTION 6

As indicated above, the vehicle-trailer system is nearly 5

linear when operating in the neighborhood of straight-

Imaginary part (rad/s)

line motion, and the corresponding state equations for a 4
4WS vehicle-trailer system can be readily written as:
3

ª v º ª a11 a12 a13 a14 º ª v º ªb11 b12 º

« r » «a a 22 a 23 a 24 »» « r » ««b21 b22 »» ªG f º 2
« » « 21 « » (3)
«I » « a 31 a 32 a 33 a 34 » «I » «b31 b32 » «¬ G r »¼ 1
« » « »« » « »
¬T ¼ ¬a 41 a 42 a 43 a 44 ¼ ¬T ¼ ¬b41 b42 ¼
0
-18 -16 -14 -12 -10 -8 -6 -4 -2 0
where I represents the hitch angle rate, and the matrix Real part (rad/s)

elements aij and bij are system parameter and speed

dependent coefficients, as listed in the Appendix. The
Figure 3: Comparison of step-input time responses of the vehicle
eigenvalue characteristics of the system can be directly alone system and the vehicle-trailer system (u=120 km/h)
derived from the system matrix in Eq. (3).
3
Impact of towing a trailer 0 Vehicle only

Vehicle yaw rate

Vehicle sideslip
Vehicle with trailer
2
Figure 2 presents a comparison of the root loci of a -5
vehicle system with and without a trailer to illustrate the 1
influence of the towed trailer on the system dynamic -10
characteristics. For the given nominal vehicle and trailer 0
0 0.5 1 1.5 0 0.5 1 1.5
system in this study, the varying parameter is the vehicle Time (sec) Time (sec)
forward speed, which increases from low to high speeds
(120 km/h). The results for low-speed portion are not 2 0
shown as stability is primarily a high-speed concern.
1
-0.2

Hitch angle
Hitch angle rate

The root-locus plots reveal an interesting distribution 0

-0.4
pattern of the vehicle-trailer system poles at relatively -1
high speeds: one pole-pair of the vehicle-trailer -2
-0.6

combination is quite comparable to that of the vehicle

-3 -0.8
without trailer case and is thus referred to as vehicle- 0 1
Time (sec)
2 3 0 1 2 3
Time (sec)
mode associated pole-pair, while the other pole-pair is
much closer to the imaginary axis and characterized by
considerably lower damping ratio and natural frequency, Influence of vehicle-trailer configuration
which is referred to as trailer-mode related pole-pair to
facilitate description. As evident from the figure, increase Compared to vehicle only, a vehicle-trailer system is
in vehicle forward speed mainly increases the magnitude subject to considerably larger parameter variations, such
of the imaginary part and reduces that of the real part of as towing different trailers with various load and load
the system poles, without changing the pole distribution distribution. These changes in system parameters, such
patterns. as mass, yaw moment of inertia, and CG location, may
significantly affect the coupling strength or interaction
Due to the impact of the trailer-related poles, the between the vehicle and the trailer, and consequently
effective damping of the vehicle-trailer system is the control design.
considerably lower as compared to vehicle alone case,
resulting in excessively oscillatory response in both hitch Figure 4 illustrates the system pole sensitivity to mass
angle and hitch angle rate, especially under low-friction variation for both vehicle only and vehicle with a trailer
surface and high speeds. In addition, due to the pole system at a high speed (120 km/h), where the masses of
deviation caused by the trailer, the vehicle also exhibits the vehicle and the trailer vary from 1000 kg to 4000 kg,
quite slower response, as shown in Figure 3. Hence, the respectively. The corresponding yaw moment of inertias
stability of the overall vehicle system is much are assumed to change proportionally with the mass
deteriorated when towing a trailer.
3
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

variation. It should be noted that the vehicle mass is kept decrease in the damping ratio, while the damping ratio of
at its nominal value while the trailer mass varies. the vehicle-associated pole-pair tends to increase. The
results thus suggest that higher mass ratio between the
System root-locus at u= 120 km/h, (m =1000~4000 kg)
2
trailer and the vehicle causes stronger coupling.
4.5
Vehicle only
Vehicle-trailer(1)
Vehicle-trailer(2) Figure 5 demonstrates the vehicle-trailer system pole
4 m2=1000 kg
sensitivity to variation in trailer CG location,
3.5 characterized by the ratio between tongue length
(distance from hitch pivot point to trailer CG) and trailer
3 length l2 (distance from hitch pivot point to trailer axle,
i.e., b2/l2).
Imaginary part (rad/s)

2.5

2
For extremely rearward trailer CG (i.e., very small b2),
m2=4000 kg
one pair of the vehicle-trailer system poles is close to
1.5 that of the vehicle only, with slightly higher damping ratio
and lower natural frequency, while the other pole-pair or
1
trailer-related pole pair is far away from the vehicle-only
0.5
pole-pair, and characterized by considerably lower
damping ratio and natural frequency, representing trailer
0
-3 -2.5 -2 -1.5 -1 -0.5 0
oscillation mode. As trailer CG moves forward, both
Real part (rad/s)
pole-pairs move away from the vehicle poles. One pole-
pair moves toward the real axis with reduced natural
Figure 4: Pole sensitivity to mass (and yaw moment of inertia) frequency and increased damping ratio, and finally
becomes two real-poles, resulting in system instability
System root-locus at vx= 70 mph, P=1.0 (b2=5%~60% L)
7
(vehicle directional stability mode) for extremely large b2
b2=60% l2
Vehicle-trailer(1) (>60%~70% l2). For b2<60% l2, with increase in b2, both
Vehicle-trailer(2)
Vehicle-trailer(3) natural frequency and damping ratio of the trailer related
6
pole-pair increase. So forward-biased trailer CG tends to
be beneficial to improving overall system damping. The
5 results thus indicate that larger trailer tongue length to
Pole of vehicle only trailer length ratio leads to weaker coupling.
Imaginary part (rad/s)

4
Evidently, to reduce the impact of the towed trailer on
b2=5% l2 the vehicle performance, desirable vehicle-trailer
3
b2=5% l2 configurations are those with large vehicle/trailer mass
ratio, low trailer yaw moment of inertia, and long trailer
2 tongue length. In practice, trailer CG is usually rearward-
biased, so the vehicle-trailer combination is inherently a
1 weekly-coupled system.
b2=60% l2
0
b2=60% l2 VEHICLE-TRAILER SYSTEM EQUILIBRIUM POINTS
-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 UNDER DIFFERENT STEERING OPERATIONS
Real part (rad/s)

Figure 5: Pole sensitivity to trailer CG location To examine the vehicle-trailer system dynamic behavior
under various steering operations, the non-linear system
It can be seen from Figure 4 that variations in trailer differential equations are further linearized at different
mass (m2) and yaw moment of inertia (Iz2) together hitch angle equilibrium points. These equilibrium points
cause both system pole-pairs to change considerably. are actually functions of front and rear wheel steering
For relatively small trailer mass (m2m1), as m2 and Iz2 angles as well as forward speed, and can be identified
increase, one pole-pair, which is comparable to the by directly solving the corresponding system steady-sate
vehicle only pole-pair in terms of damping ratio and thus equations of motion. Analytical solutions to the steady-
can be referred to as vehicle-associated pole-pair, state equations, though readily available, tend to be
departs from the vehicle only pole-pair and drops quite lengthy, so numerical results are presented in this
considerably in natural frequency with moderate paper.
increase in damping ratio. The other pole-pair, which is
characterized by a considerably lower damping ratio and Figure 6 presents the vehicle-trailer equilibrium point
a relatively higher natural frequency and thus can be distribution as functions of front and rear wheel steering
considered trailer-associated pole-pair, moves toward inputs at high speed. The front and rear wheel steer
the imaginary axis. Further increase in trailer mass angles are varied in a large range from 0 to 10 degrees,
(m2!m1) causes the nature frequency of the trailer respectively. Figure 7 shows the equilibrium point
related pole-pair to drop considerably with only slight distribution as functions of vehicle speed for different
4
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

vehicle front wheel steering angle inputs, while the quite small, indicating that the operating point tends to
vehicle rear wheel is kept straight (i.e., two-wheel influence the system dynamics only slightly. In the case
steering or 2WS). of 2WS vehicle-trailer combination (Gr=0), as front wheel
Vehicle-trailer equilibrium point distribution (vx =120 kph) steering angle increases, the real parts of the two pole-
25 20 pairs depart from each other, while their imaginary parts
20 tend to become closer. When both front and rear wheels
10
are steered (4WS) in phase, however, an opposite

Yaw rate (deg./s)

Sideslip (kph)

15

10 0 tendency occurs as front wheel steering angle

5 increases. Due to the offset of the two oppositely moving
0
-10
system pole-pairs, the overall influence of operating
-5 -20
point variation can thus be considered relatively
0 2 4 6
Front wheel steer (deg.)
8 10 0 2 4 6
Front wheel steer (deg.)
8 10
insignificant.
15
Vehicle-trailer pole distribution (vx =120 kph)
10 Gr=0 6
Hitch angle (deg.)

Gr=2
Gr=4 Gf =0~10 deg.
5

Imaginary part
Gr=6 5.5
Gr=8 G r=0
0 G r=4
Gr=10
G r=8
5
-5
0 2 4 6 8 10
Front wheel steer (deg.)
4.5
-5 -4.8 -4.6 -4.4 -4.2 -4 -3.8 -3.6 -3.4 -3.2 -3
Figure 6: Vehicle-trailer equilibrium point distribution as Real part (poles 1 and 2)
functions of front and rear wheel steering inputs
Vehicle-trailer equilibrium point distribution (Gr=0) 6
2 25 Imaginary part

20 5.5
0
Yaw rate (deg./s)
Sideslip (kph)

15
-2
10 5 Gf =0~10 deg

-4
5
4.5
-6 0 -2.45 -2.4 -2.35 -2.3 -2.25 -2.2 -2.15 -2.1 -2.05
20 40 60 80 100 120 20 40 60 80 100 120 Real part (poles 3 and 4)
Speed (kph) Speed (kph)
Figure 8: Influence of operation points on the poles of the
0 vehicle-trailer system

Magnitude (dB)
20 10
-5 Gf =0 0
Hitch angle (deg.)

10 -10
Gf =2
Trailer yaw rate

0 Hitch angle -20

-10 Gf =4 -30
-10
Gf =6 0
-180

Phase (deg)
Gf =8
-15
Gf =10
-180
-20 -360
20 40 60 80 100 120 0 1 0 1
10 10 10 10
Speed (kph) Frequency (rad/sec) Frequency (rad/sec)

Figure 7: Vehicle-trailer equilibrium point distribution as

functions of speed and front wheel steering input (Gr=0) 10 0
Trailer yaw rate (deg./sec.)

Gf0=0.
8 Gf0=10 deg. -0.5
Hitch angle (deg.)

The figures reveal that the magnitude of the steady-state 6

-1

hitch angle decreases with increasing vehicle speed. At 4

-1.5
higher speed, the magnitude of the steady-state hitch 2
-2
angle value tends to be small although the transient 0

magnitude may be large. The above results thus show -2

0 1 2 3
-2.5
0 1 2 3
Time (sec) Time (sec)
that, in terms of steady-state response, the hitch angle
can be reasonably considered small, especially at Figure 9: Frequency and step-input time response of the vehicle-
relatively high speed. trailer at two operating points (Gr0=10 deg., 'Gf=3 deg)

INFLUENCE OF HITCH ANGLE OPERATING POINTS Figure 9 presents a comparison of the frequency and
time response of the vehicle-trailer system linearized at
Figure 8 illustrates the influence of operating points on two different operating points, corresponding to both
the poles of the vehicle-trailer system at high speed. The small and large steering inputs, respectively, at 120
plot reveals that, for a given speed and rear wheel km/h. The results further demonstrate that the influence
steering angle, the two pole-pairs of the system move in due to operating point variation is quite small.
opposite directions. The varying ranges, however, are
5
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

To verify the effect of operating points at large hitch CONTROL OBJECTIVES

angle, further simulation using VehSim [8] is conducted.
A step steer of 2700 is applied at a low speed, which is
to exclude the nonlinearity caused by other factors, such From (3), a reference vehicle equipped with 4WS open
as the tire. The steady state hitch angle is around 280. loop control G rop can be modeled linearly through a
Figure 10 shows that the response from linear model second-order bicycle model (assuming high-co surface,
matches very well with the nonlinear full vehicle model. small steering and linear tire model):

ªv º ª a11 a12 º ªv º ªb11 º ªb º (4)

«r » «a » « »  « »G f  « 12 »G rop
¬ ¼ ¬ 21 a 22 ¼ ¬r ¼ ¬b21 ¼ ¬b22 ¼

or,

ªvº ª a11 a12 º ªv º ª b11  K f b12 º (5)

«a « »G f
«r»
¬ ¼ ¬ 21 a 22 »¼ «¬r »¼ ¬b21  K f b22 ¼

G rop
where K f is the ratio between rear-wheel steer
Gf
and front-wheel steer, which is designed to ensure
steady-state zero sideslip, and the coefficients, aij and
bij, are similarly defined as in the Appendix, but with
vehicle only. The desirable sideslip and yaw rate
characteristics can be derived from (5), and given by:

( b 11 K f b 12 )s
v d (s ) = 2
s (a 11 + a 22 )s + (a 11 a 22 a 12 a 21 )

(b21  K f b22 )s  K f (a11b22  a21b12 )  a21b11  a11b21

rd ( s) (6)
s 2  (a11  a22 )s  (a11a22  a12 a21 )

And from (3), the steady-state hitch angle under 4WS

open-loop control can be derived as:

ș dss = f (u )į f (7)

where f(u) represents the vehicle and trailer geometric

and dynamic parameters and is a function of vehicle
forward speed.

An optimal control method is used with the performance

index defined below:
Figure 10: Vehicle yaw rate and hitch angle responses at a low
T
speed
J ³ [I
2
 w1 (T  T dss ) 2  w2 ( v  v d ) 2  w3 ( r  rd ) 2 ]dt (8)
0
CONTROLLER DESIGN CONSIDERATIONS

Many control methods can be utilized to control the where wi (i=1, 2 and 3) is the weight factor, while vd and
vehicle-trailer system, such as active steering and/or rd are the desired performance of the vehicle alone to
differential braking control [5-7]. In this paper, a state- preserve the driver's perception.
feedback controller is designed, based on partial or full
state feedback with the state-space equations described The objective of the state-feedback control is to track the
by (3), where the rear-wheel steering angle is taken as desired performance of the vehicle-trailer combination,
the control input. that is, to preserve the driver’s perception when driving
the vehicle alone, and, in the meantime, to stabilize the
trailer by minimizing the hitch angle rate and driving the
hitch angle to the desired one defined in (7).

6
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

CONTROL SCHEMES
4 1.8

From (3), 3.5 1.6

3
1.4
Vehicle reference

x Ax  B1G f  B2G r (9)

Vehicle yaw rate (deg/sec)

Vehicle sideslip (m/sec)
2.5 Feedback from vehicle
Feedback from trailer 1.2
2 Full state feedback

where B1 and B2 are, respectively, the first and second

1
1.5
0.8
column vectors of the input matrix B. 1
0.6
0.5
0.4
0
Depending upon the situations in practice, such as -0.5 0.2

availability of sensors on vehicle and/or on trailer, three -1

0 0.5 1 1.5 2 2.5
0
0 0.5 1 1.5 2 2.5

types of state-feedback control law may be adopted with Time (sec) Time (sec)

either partial or full state feedback. 1.5 0.1

1 0

Vehicle state feedback 0.5 -0.1

Hitch angle rate (deg/sec)

Hitch angle (deg)

0 -0.2

Gr k1v  k 2 r (10) -0.5 -0.3

-1 -0.4

where k1 and k2 are sideslip and yaw rate feedback -1.5 -0.5

gains, respectively. -2 -0.6

It is well known that a 4WS vehicle can stabilize the -2.5

0.5 1 1.5 2 2.5
-0.7
0.5 1 1.5 2 2.5

vehicle-trailer combination to some extent. Owing to the Time (sec) Time (sec)

fact that one of the vehicle state variables is practically Figure 11: Vehicle-trailer time response under different controls
immeasurable, an alternative approach, such as state 5
120 km/h
estimation, needs to be taken. This is in fact the most 4.5
practical means since no sensor is required from the 50 km/h
trailer. However, as can be seen from Figure 11, the 4

trailer stability shows very small improvement. Figure 12 3.5

presents the system pole distribution under vehicle 120 km/h
60 km/h
state-feedback control as expressed in (10). The root-
Imaginary part (rad/s)

locus plot shows that, with vehicle state-feedback 2.5

control, the pole-pair mainly associated with the vehicle 120 km/h
is always highly damped at relatively high speed. The 2

existence of the trailer shifts the vehicle poles towards 1.5

50 km/h

the imaginary axis, as shown in the figure, similar to the

base vehicle-trailer case. The poles primarily associated 1
Vehicle only

with the trailer, however, are only slightly affected by the 0.5
Vehicle with trailer (1)
Vehicle with trailer (2)

vehicle state-feedback control, due to the relatively weak

coupling between the vehicle and the trailer, especially 0
-11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0
when the loading is close to the trailer axle. With vehicle Real part (rad/s)

state-feedback, the response of the vehicle with the Figure 12: System pole distribution under vehicle state-feedback
trailer moderately departs from that of the vehicle only System poles with hitch angle rate feedback (vx =5~120 km/h, P=1.0)
case. The hitch angle and hitch angle rate response, 5
though considerably enhanced through the vehicle
4.5
control (with reduced peak and steady-state values), still
exhibit excessive oscillation, especially at high speeds. 4

Trailer state feedback 3.5

pole 1
Imaginary part (rad/s)

pole 2
Gr k 3I  k 4T (11) 3
pole 3

where k3 and k4 are hitch angle rate and hitch angle 2.5

feedback gains, respectively. Trailer state feedback is

2
like adding a damper (and a spring) onto the rotating
hitch pivot between the vehicle and the trailer, leading to 1.5 120 km/h
significantly increased coupling. As expected, it
stabilizes the trailer very effectively. However, this 1

control law may change the vehicle characteristics 0.5

dramatically, as demonstrated by the system time Pole 1 at 120 km/h40 km/h Pole 3 at 120 km/h
response in Figure 11 and the mixed system pole 0
-11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0
patterns in Figure 13, which may impose considerable Real part (rad/s)

discomfort to the driver. Figure 13: System pole distribution under hitch angle rate feedback

7
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

Full state feedback 4. El-Gindy, M., Directional Response of a Tractor

Towing a Semi-Trailer, Int. J. of Vehicle Design, Vol.
Gr k1 v  k 2 r  k 3I  k 4T (12) 10, No. 2, 1989, pp. 210-226
5. Kimbrough, S., Elwell, M. and Chiu, C., Braking
Although this approach requires either measurement or Controllers and Steering Controllers for Combination
estimation on all state variables, it provides the optimal Vehicles, Heavy Vehicle Systems, Int. J. of Vehicle
way to effectively stabilize the vehicle-trailer combination Design, Vol. 1, No. 2, 1994, pp. 195-223
while preserving the driver’s perception when driving the 6. Kageyama, I. and Nagai, R., Stabilization of
vehicle alone, as illustrated in Figure 11. Passenger Car-Caravan Combination using Four
Wheel Steering Control, Vehicle System Dynamics,
CONCLUSIONS Vol. 24, 1995, pp. 313-327
7. Lugner, P., Plochl, M. and Riepl, A, Investigation of
It is observed from this study that the coupling between Passenger Car-Trailer Dynamics Controlled by
vehicle and trailer as two pivoted rigid bodies is Additional Braking of the Trailer, Proc. of AVEC'96,
inherently weak and its strength depends largely on the Aachen, 1996
ratio of the masses and their distribution of the two- 8. W. Deng and et al, Fast Software Prototype
connected bodies. An optimal control method with state- Development for Chassis/Driveline Controls and
feedback rear-wheel steer is applied, which further Integration, Int. J. of Vehicle Design, Vol. 29, 2001,
demonstrates that in order to improve vehicle-trailer pp. 139-147
stability while preserving driver’s perception, feedback
with states from both vehicle and trailer is necessary,
although the emphasis can be varied depending on the
objective.
APPENDIX
The variation of hitch angle operating points has been a
concern from linear analysis and control design point of SIGN CONVENTIONS
view. However, it reveals relatively insignificant effect on
the system stability characteristics.
x Vehicle body coordinate system: the longitudinal and
lateral axes (ox and oy) point to vehicle heading
REFERENCES direction and vehicle left side, respectively, as
shown in Figure 1.
1. Jindra, F., Tractor and Trailer Handling, Automobile x Yaw motion: positive when rotating counterclockwise
Engineer, Vol. 55, No. 2, 1965, pp. 60-69 from the longitudinal axis to the lateral axis, as
2. Bundorf, R.T., Directional Control Dynamics of shown in Figure 1.
Automobile-Travel Trailer Combinations, SAE Paper x Hitch angle: positive when rotating counterclockwise
No.: 670099, 1967 from the vehicle longitudinal axis to the trailer
3. Huston, J.C. and Johnson, D.B., Basic Analytical longitudinal axis.
Results for Lateral Stability of Car/Trailer Systems,
SAE Paper No.: 820136, 1982

8
 Downloaded from SAE International by Univ of California Berkeley, Saturday, August 04, 2018

EQUATIONS OF MOTION OF AN ALL-WHEEL STEERING VEHICLE-TRAILER SYSTEM

ª m1  m2  m2 (c  a2 cosT )  m2 a2 cosT º ­ v ½
« m1c I z1 0 » ° r °
« »® ¾
«¬ m2 a2 cosT 2
I z 2  m2 a2  m2 ca2 cosT I z 2  m2 a22 »¼ °¯T°¿
ª º
«  (CDf  CDr  CDt cosT )  a1CDf  b1CDr  (c cosT  a2  b2 )CDt  (m1  m2 )u 2 (a2  b2 )CDt » ­ v ½
1« »° ° (A1)
 (a1  c)CDf  (b1  c)CDr  a1 (a1  c)CDf  b1 (c  b1 )CDr  m1cu 2 0 »®r ¾
u«
« (a  b )(c cosT  a2  b2 ) (a  b ) 2
» °T°
« (a2  b2 )CDt  2 2 CDt  m2 a2u 2 cosT  2 2 CDt » ¯ ¿
¬ cosT cosT ¼
ª CDf CDr CDt cosT º ­ G f ½
« » ° °
 «(a1  c)CDf (c  b1 )CDr 0 »® G r ¾
«¬ 0 0  (a2  b2 )CDt ¼ ¯G t  tan T °¿
» °

COEFFICIENTS IN EQUATION (3)

A11 A1
a11  (CDf  CDr  CDt ), a12  1 [a1CDf  b1CDr  (c  a2  b2 )CDt  (m1  m2 )u2 ]
u u
1 1
A1 A1 A11
a13 (a2  b2 )CDt , a14 CDt , a21 [(b1  c)CDr  (a1  c)CDf ]
u u u
A1
a22  1 [a1(a1  c)CDf  b1(c  b1)CDr  m1cu2 ], a23 0, a24 0
u
1 (A2)
A1 A11
a31 (a2  b2 )CDt , a32 [m2a2u2  (a2  b2 )(c  a2  b2 )CDt ]
u u
A1 A1
a33  1 (a2  b2 )2 CDt , a34  1 (a2  b2 )CDt , a41 0, a42 0, a43 A11, a44 0
u u
1 1
b11 A1 CDf , b21 A1 (a1  c)CDf , b31 0, b41 0
b12 A11CDr , b22 A11(c  b1)CDr , b32 0, b42 0

ªm1  m2  m2 (c  a2 )  m2a2 0º
« mc I z1 0 0»»
where A1 « 1 (A3)
«  m2a2 I z 2  m2a2  m2ca2 I z 2  m2a22
2
0»
« »
¬ 0 0 0 1¼

and CDf, CDr and CDt represent cornering stiffness of tires at vehicle front, vehicle rear and trailer axle, respectively.